From l.karssen at erasmusmc.nl Tue Mar 1 09:16:15 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Tue, 01 Mar 2011 09:16:15 +0100 Subject: [GenABEL-dev] [Genabel-commits] r665 - pkg/webpages In-Reply-To: References: <20110228172335.2F9FE9CCF0@r-forge.r-project.org> Message-ID: <1298967375.3908.11.camel@barabas> Hi Yurii, I took a quick look at the tutorial at the URL you gave below. Looks good! Looking at the HTML code I wondered: I see a lot of

lines. IIRC the idea of the

code is to create a paragraph. I guess

gives you a separation between paragraphs as well, but I think it was meant to be used like this:

A lot of text in a paragraph

Keep in mind though that my HTML skills are pre-2000 :-), so check somewhere else as well. Furthermore, I think that drupal knows the tage so you might want to consider enclosing lines like R_HOME= etc. in these tags for nice source code formatting. Lennart. On Mon, 2011-02-28 at 18:29 +0100, Yurii Aulchenko wrote: > ok, here is "How to install devel-version of GenABEL suite" (plain HTML) > > also available at > > https://r-forge.r-project.org/scm/viewvc.php/pkg/webpages/tutHowToInstallDevelVersion.html?view=markup&root=genabel > > Any further comments appreciated. Let us have few more days for anyone > to comment, and then if no issues pop up, will move this the web-site > tutorials section. > > My idea was to put this on genabel.org/tutorials, as this is now > something for 'advanced user', not a developer. > > Yurii > > > +

> > +How to install devel-version of GenABEL suite > > +

> > +

> > + > > +by Yurii Aulchenko, William Astle and Lennart Karssen for the GenABEL-devel team > > + > > +

> > + > > +Here are instructions on how to get devel-version of GenABEL packages > > +working, assuming you use Linux or Mac OS. > > +

> > + > > +To get GenABEL & Co. running in devel-setting, you need to do a number of > > +things. Plan to spend about 30 minutes on that. You will need good > > +Inet connection and you also will > > +need to have Subversion (svn) installed. > > +

> > + > > +1) Check if Subversion is installed. For that, try 'svn help'. If you > > +do get help, everyting must be all right. If Subversion is not there, > > +here is the link, which may help: > > + > > +http://subversion.apache.org/packages.html > > + > > +

> > + > > +2) In order to be able to compile devel-dersion, you need to set > > +an environmental variable R_HOME to point the location, where > > +R is installed. We assume that you use Bash (most people do); if not, > > +try to figure out how you can set environmental variables for your particular environment (and let us know so we can add it here :) ). > > +

> > + > > +You can check where R is installed by running 'dirname $(which R)' > > +on Bash command prompt. Assuming R is installed at '/usr/bin', add the following lines to your ~/.bashrc file > > +

> > + > > +R_HOME=/usr/bin > > +
> > +export R_HOME > > +

> > + > > +or, you can try > > +

> > + > > +R_HOME=$(dirname $(which R)) > > +
> > +export R_HOME > > +

> > + > > + > > +log out and log in again, or start a new terminal/console window. > > +

> > + > > +3) Install _stable_ GenABEL et al., 'roxygen', and 'RUnit' > > +from CRAN; for that on R command line type > > +

> > + > > +install.packages("GenABEL",dependencies=c("Depends","Suggests")) > > +
> > +install.packages("roxygen") > > +
> > +install.packages("RUnit") > > +

> > + > > +3) Change to a place, where you want the source code > > +to go. Get to the latest code by > > +

> > + > > +svn export svn://svn.r-forge.r-project.org/svnroot/genabel/pkg > > +

> > + > > +This should create directory called 'pkg' with all packages > > +in it. > > +

> > + > > +4) Compile and install latest code by > > +

> > + > > +cd pkg > > +
> > +R CMD INSTALL DatABEL > > +
> > +R CMD INSTALL GenABEL > > +
> > +... etc. > > +

> > + > > +Note some packages (e.g. ProbABEL) are not R packages; to > > +compile these use specific Makefile's, e.g. > > +

> > +cd ProbABEL > > +
> > +make > > +

> > + > > +FINITO! Enjoy :) > > +

> _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room 2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 e-mail: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From william.astle at imperial.ac.uk Tue Mar 1 11:51:32 2011 From: william.astle at imperial.ac.uk (William Astle) Date: Tue, 01 Mar 2011 10:51:32 +0000 Subject: [GenABEL-dev] faster polygenic In-Reply-To: References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se> <4D6B7C98.60009@ic.ac.uk> Message-ID: <4D6CCFB4.1060907@ic.ac.uk> Hi On 28/02/11 21:36, Yurii Aulchenko wrote: > 2) still, FMM generates L which are>1 (1.014), and this is VERY > significant. Old and hglm-based tend to be a bit on other side -- L< > 1 (0.996 for old, 0.993 for hglm), and for hglm this is more > pronounced That's interesting, I guess the extra variance inflation is probably due to uncertainty in K which we don't model. I've wondered about extending the method to work with an uncertain K, I'm not sure how feasible statistically or computationally that would be. bw Will From william.astle at imperial.ac.uk Tue Mar 1 11:55:52 2011 From: william.astle at imperial.ac.uk (William Astle) Date: Tue, 01 Mar 2011 10:55:52 +0000 Subject: [GenABEL-dev] [Genabel-commits] r665 - pkg/webpages In-Reply-To: <1298967375.3908.11.camel@barabas> References: <20110228172335.2F9FE9CCF0@r-forge.r-project.org> <1298967375.3908.11.camel@barabas> Message-ID: <4D6CD0B8.2060701@ic.ac.uk> Hi Yurii, There is a typo: 2) In order to be able to compile devel-dersion, 'v'->'d' bw Will From yurii.aulchenko at gmail.com Wed Mar 2 16:15:24 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 2 Mar 2011 16:15:24 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: <4D6CCFB4.1060907@ic.ac.uk> References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se> <4D6B7C98.60009@ic.ac.uk> <4D6CCFB4.1060907@ic.ac.uk> Message-ID: >> 2) still, FMM generates L which are>1 (1.014), and this is VERY >> significant. Old and hglm-based tend to be a bit on other side -- L< >> 1 (0.996 for old, 0.993 for hglm), and for hglm this is more >> pronounced > > That's interesting, I guess the extra variance inflation is probably due to > uncertainty in K which we don't model. I've wondered about extending the > method to work with an uncertain K, I'm not sure how feasible statistically > or computationally that would be. This is very interesting suggestion -- never quite thought of what woul the effects of uncertainty of K be onto test statistics. My feeling is that uncertainty in K is likely to translate into uncertainty of Lambda (so, SD of L would be bigger), but not into bias -- and that is what we see; but I can well be wrong about that. I think we can get a better view of the problem and answer to William's hypothesis when we get polygenic-FMM working as an entity, separate from FMM-GWAS and generating standard 'null polygenic' outputs. Now we kind of compare apples and pears by contrasting old/hglm-null-polygenic + MMSCORE-GWAS with null-polygenic-FMM + FMM-GWAS, and FMM-GWAS as a method is different from MMSCORE. So I really wonder what would come out if we generate null-polygenic with all three methods and can push the GWAS later on with both MMSCORE and FMM. I believe we may have important clues on the above formulated problem from that comparison. best wishes, Yurii From william.astle at imperial.ac.uk Wed Mar 2 16:59:13 2011 From: william.astle at imperial.ac.uk (William Astle) Date: Wed, 02 Mar 2011 15:59:13 +0000 Subject: [GenABEL-dev] faster polygenic In-Reply-To: References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se> <4D6B7C98.60009@ic.ac.uk> <4D6CCFB4.1060907@ic.ac.uk> Message-ID: <4D6E6951.8090709@ic.ac.uk> On 02/03/11 15:15, Yurii Aulchenko wrote: >> That's interesting, I guess the extra variance inflation is probably due to >> uncertainty in K which we don't model. I've wondered about extending the >> method to work with an uncertain K, I'm not sure how feasible statistically >> or computationally that would be. > This is very interesting suggestion -- never quite thought of what > woul the effects of uncertainty of K be onto test statistics. My > feeling is that uncertainty in K is likely to translate into > uncertainty of Lambda (so, SD of L would be bigger), but not into bias > -- and that is what we see; but I can well be wrong about that. > You might be right, I've not thought a lot about it, but my initial thinking was there might be some bias, my line of thought was: if there is maximal uncertainty in K then K adds no useful information, so you'd expect to estimate heritability=0, but then your statistics are just uncorrected chisq, so you have a standard GC situation in which case lambda>1. However if K is accurate you'd hope that the statistic is correct therefore lambda=1. But maybe this is not quite what you mean by bias? > I think we can get a better view of the problem and answer to > William's hypothesis when we get polygenic-FMM working as an entity, OK, I'll see what I can do. Is the best approach to have the user create a Polygenic-FMM R object and then let the user apply an R function to it when they want to do the GWAS? GWAS-FMM(Polygenic-FMM-Object, GWAS-SNP-Data) In this case we will need to call two separate C functions from R, when to setup and fit polygenic-FMM object and then another to run the GWAS. bw Will From yurii.aulchenko at gmail.com Wed Mar 2 18:04:00 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 2 Mar 2011 18:04:00 +0100 Subject: [GenABEL-dev] [Genabel-commits] r665 - pkg/webpages In-Reply-To: <1298967375.3908.11.camel@barabas> References: <20110228172335.2F9FE9CCF0@r-forge.r-project.org> <1298967375.3908.11.camel@barabas> Message-ID: > Looking at the HTML code I wondered: I see a lot of

lines. IIRC > the idea of the

code is to create a paragraph. I guess

gives > you a separation between paragraphs as well, but I think it was meant to > be used like this: >

> A lot of text in a paragraph >

This is correct. But

generated the same pagination, and was easier to do (and maintain later, I believe). I do not see a problem to use this style, but would also be OK with "right style". > Furthermore, I think that drupal knows the tage so you > might want to consider enclosing lines like R_HOME= etc. in these tags > for nice source code formatting. This is good point. Big question hidden here: do we want our 'html' code be Drupal-dependent or not? I say, rather not, and would go for use of

instead as a standard way to get verbatim, unless someone suggest otherwise. bw, Yurii From yurii.aulchenko at gmail.com Thu Mar 3 10:11:52 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 3 Mar 2011 10:11:52 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: <4D6E6951.8090709@ic.ac.uk> References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se> <4D6B7C98.60009@ic.ac.uk> <4D6CCFB4.1060907@ic.ac.uk> <4D6E6951.8090709@ic.ac.uk> Message-ID: >>> That's interesting, I guess the extra variance inflation is probably due >>> to >>> uncertainty in K which we don't model. I've wondered about extending the >>> method to work with an uncertain K, I'm not sure how feasible >>> statistically >>> or computationally that would be. >> >> This is very interesting suggestion -- never quite thought of what >> woul the effects of uncertainty of K be onto test statistics. My >> feeling is that uncertainty in K is likely to translate into >> uncertainty of Lambda (so, SD of L would be bigger), but not into bias >> -- and that is what we see; but I can well be wrong about that. >> > > You might be right, I've not thought a lot about it, but my initial thinking > was there might be some bias, my line of thought was: if there is maximal > uncertainty in K then K adds no useful information, so you'd expect to > estimate heritability=0, but then your statistics are just uncorrected > chisq, so you have a standard GC situation in which case lambda>1. However > if K is accurate you'd hope that the statistic is correct therefore > lambda=1. I see your point. This is an interesting 'extreme' scenario to consider. Would be really interesting to try that: say, use less and less markers to estimate K and see what happens. Rather trivial to set up, and may give us important clues on how many markers should be used for K. I was thinking from under an assumption that even in population-based studies you can correct for inflation (Lambda>1) by use of K. This is actually another interesting question to test using real data :) But you are right, probably you need very good K to do that. > But maybe this is not quite what you mean by bias? > By 'bias' I indeed mean deviation of Lambda from 1. >> I think we can get a better view of the problem and answer to >> William's hypothesis when we get polygenic-FMM working as an entity, > > OK, I'll see what I can do. Is the best approach to have the user create a > Polygenic-FMM R object and then let the user apply an R function to it when > they want to do the GWAS? > > GWAS-FMM(Polygenic-FMM-Object, GWAS-SNP-Data) > > In this case we will need to call two separate C functions from R, when to > setup and fit polygenic-FMM object and then another to run the GWAS. > Absolutely, this was what I though about. Polygenic-FMM-Object should comply to 'polygenic' format -- meaning it should have a few slots named in particular way, but not necessarily all, and can contain extra slots with info not in standard polygenic (e.g. I remember you derive something big during null-herit estimation, which is not in standard polygenic). Then we could thinking of coming up with 'consolidated' polygenic-class at the end, or decide this is not worth effort :) all the best, Yurii From william.astle at imperial.ac.uk Thu Mar 3 11:36:01 2011 From: william.astle at imperial.ac.uk (William Astle) Date: Thu, 03 Mar 2011 10:36:01 +0000 Subject: [GenABEL-dev] faster polygenic In-Reply-To: References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se> <4D6B7C98.60009@ic.ac.uk> <4D6CCFB4.1060907@ic.ac.uk> <4D6E6951.8090709@ic.ac.uk> Message-ID: <4D6F6F11.9080305@ic.ac.uk> On 03/03/11 09:11, Yurii Aulchenko wrote: > > Absolutely, this was what I though about. Polygenic-FMM-Object should > comply to 'polygenic' format -- meaning it should have a few slots > named in particular way, but not necessarily all, and can contain > extra slots with info not in standard polygenic (e.g. I remember you > derive something big during null-herit estimation, which is not in > standard polygenic). Then we could thinking of coming up with > 'consolidated' polygenic-class at the end, or decide this is not worth > effort :) great - I will look at polygenic-class to see what I need to add Thanks for this info Will From yurii.aulchenko at gmail.com Thu Mar 3 14:37:55 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 3 Mar 2011 14:37:55 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: <4D6F6F11.9080305@ic.ac.uk> References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se> <4D6B7C98.60009@ic.ac.uk> <4D6CCFB4.1060907@ic.ac.uk> <4D6E6951.8090709@ic.ac.uk> <4D6F6F11.9080305@ic.ac.uk> Message-ID: >> Absolutely, this was what I though about. Polygenic-FMM-Object should >> comply to 'polygenic' format -- meaning it should have a few slots >> named in particular way, but not necessarily all, and can contain >> extra slots with info not in standard polygenic (e.g. I remember you >> derive something big during null-herit estimation, which is not in >> standard polygenic). Then we could thinking of coming up with >> 'consolidated' polygenic-class at the end, or decide this is not worth >> effort :) > > great - I will look at polygenic-class to see what I need to add > What may be a good plan -- if you make fixed effects estimates, and h2 + total var (or anything which can help computing these) visible, we already can use that for mmscore -- along the lines with 'polygenic_hglm' https://r-forge.r-project.org/scm/viewvc.php/pkg/GenABEL/R/polygenic_hglm.R?view=markup&revision=663&root=genabel where I recover all other staff from these parameters (see lines after the call "res_hglm <- hglm(y = y, X = desmat, Z = L, family = family, ... )", where I fake much of the 'polygenic' object from these params). Surely, most of the things you already compute inside, and eventually it all can be done better and much faster; but why not having something working and run some comparison already. bw, Yurii From william.astle at imperial.ac.uk Thu Mar 3 14:41:29 2011 From: william.astle at imperial.ac.uk (William Astle) Date: Thu, 03 Mar 2011 13:41:29 +0000 Subject: [GenABEL-dev] faster polygenic In-Reply-To: References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se> <4D6B7C98.60009@ic.ac.uk> <4D6CCFB4.1060907@ic.ac.uk> <4D6E6951.8090709@ic.ac.uk> <4D6F6F11.9080305@ic.ac.uk> Message-ID: <4D6F9A89.5090003@ic.ac.uk> Yurii wrote: > Surely, most of the things you already compute inside, and eventually > it all can be done better and much faster; but why not having > something working and run some comparison already. okay - i will se.e what i can do, bw Willl From yurii.aulchenko at gmail.com Fri Mar 4 22:48:01 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Fri, 4 Mar 2011 22:48:01 +0100 Subject: [GenABEL-dev] [Genabel-commits] r666 - pkg/webpages In-Reply-To: <20110304184415.BBDBD9CD80@r-forge.r-project.org> References: <20110304184415.BBDBD9CD80@r-forge.r-project.org> Message-ID: Modified tutHowToInstallDevelVersion along the lines suggested by Lennart and William. Now this is ready to go into Tutorials section of the web-site Yurii On Fri, Mar 4, 2011 at 7:44 PM, wrote: > Author: yurii > Date: 2011-03-04 19:44:15 +0100 (Fri, 04 Mar 2011) > New Revision: 666 > > Modified: > ? pkg/webpages/tutHowToInstallDevelVersion.html > Log: > webpages how to install dev version -- few changes suggested in discussion > > Modified: pkg/webpages/tutHowToInstallDevelVersion.html > =================================================================== > --- pkg/webpages/tutHowToInstallDevelVersion.html ? ? ? 2011-02-28 17:23:34 UTC (rev 665) > +++ pkg/webpages/tutHowToInstallDevelVersion.html ? ? ? 2011-03-04 18:44:15 UTC (rev 666) > @@ -25,7 +25,7 @@ > ? > ?

> > -2) In order to be able to compile devel-dersion, you need to set > +2) In order to be able to compile devel-version, you need to set > ?an environmental variable R_HOME to point the location, where > ?R is installed. We assume that you use Bash (most people do); if not, > ?try to figure out how you can set environmental variables for your particular environment (and let us know so we can add it here :) ). > @@ -35,17 +35,19 @@ > ?on Bash command prompt. Assuming R is installed at '/usr/bin', add the following lines to your ~/.bashrc file > ?

> > +

> ?R_HOME=/usr/bin
> -

> ?export R_HOME
> +

> ?

> > ?or, you can try > ?

> > +

> ?R_HOME=$(dirname $(which R))
> -

> ?export R_HOME
> +

> ?

> > > @@ -56,18 +58,20 @@ > ?from CRAN; for that on R command line type > ?

> > +

> ?install.packages("GenABEL",dependencies=c("Depends","Suggests"))
> -

> ?install.packages("roxygen")
> -

> ?install.packages("RUnit")
> +

> ?

> > ?3) Change to a place, where you want the source code > ?to go. Get to the latest code by > ?

> > +

> ?svn export svn://svn.r-forge.r-project.org/svnroot/genabel/pkg
> +

> ?

> > ?This should create directory called 'pkg' with all packages > @@ -77,21 +81,22 @@ > ?4) Compile and install latest code by > ?

> > +

> ?cd pkg
> -

> ?R CMD INSTALL DatABEL
> -

> ?R CMD INSTALL GenABEL
> -

> ?... etc.
> +

> ?

> > ?Note some packages (e.g. ProbABEL) are not R packages; to > ?compile these use specific Makefile's, e.g. > ?

> + > +

> ?cd ProbABEL
> -

> ?make
> +

> ?

> > ?FINITO! Enjoy :) > > _______________________________________________ > Genabel-commits mailing list > Genabel-commits at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits > From yurii.aulchenko at gmail.com Sun Mar 6 20:57:41 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Sun, 6 Mar 2011 20:57:41 +0100 Subject: [GenABEL-dev] [Genabel-commits] r667 - in pkg/GenABEL: R man src/GAlib In-Reply-To: <20110306191118.1FFF59CE46@r-forge.r-project.org> References: <20110306191118.1FFF59CE46@r-forge.r-project.org> Message-ID: Dear All, In this commit I have addressed the feature request "option to have header line missing in map-files used by convert.snp.ped" https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1317&group_id=505&atid=2061 for reason why, see on forum http://forum.genabel.org/viewtopic.php?f=6&t=14#p197 This appears to be very straightforward thing to do, but please have a look. I did simple test using data in pkg/GenABEL/inst/exdata with and without header line included in map.18, and all looked fine (resulting .raw's diff = no diff). You will see one more related commit on the commit-list -- I am sorry about that, I was breaking the guideline not to commit before doing R CMD check, and a couple of minor things (discrepancy documentation/R-code) popped up there... Never again! Yurii On Sun, Mar 6, 2011 at 8:11 PM, wrote: > Author: yurii > Date: 2011-03-06 20:11:17 +0100 (Sun, 06 Mar 2011) > New Revision: 667 > > Modified: > ? pkg/GenABEL/R/convert.snp.ped.R > ? pkg/GenABEL/man/convert.snp.ped.Rd > ? pkg/GenABEL/src/GAlib/convert_snp_merlin.cpp > ? pkg/GenABEL/src/GAlib/convert_snp_merlin_wslash.cpp > Log: > introduced an option 'mapHasHeaderLine' in convert.snp.ped, addressing feature request [#1317] > > Modified: pkg/GenABEL/R/convert.snp.ped.R > =================================================================== > --- pkg/GenABEL/R/convert.snp.ped.R ? ? 2011-03-04 18:44:15 UTC (rev 666) > +++ pkg/GenABEL/R/convert.snp.ped.R ? ? 2011-03-06 19:11:17 UTC (rev 667) > @@ -1,29 +1,40 @@ > -"convert.snp.ped" <- function(pedfile,mapfile,outfile,format="premakeped",traits=1,strand="u",bcast=10000000,wslash=F) { > - > +"convert.snp.ped" <- > + ? ? ? ? ? ? ? function(pedfile,mapfile,outfile,format="premakeped",traits=1, > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? strand="u",bcast=10000000,wslash=FALSE,mapHasHeaderLine=TRUE) > +{ > + > ? ? ? ?alcodes <- alleleID.codes() > - > + > ? ? ? ?poss <- c("u","+","-","file") > ? ? ? ?intstrand <- which(poss == strand)-1 > ? ? ? ?if (length(intstrand)!=1) {cat("strand argument must be one of",poss,"\n");stop();} > - > + > ?# strand == 0 -> unknown ('u') > ?# strand == 1 -> plus ('+') > ?# strand == 2 -> minus ('-') > ?# extendedmap -> five columns (chr,name,pos,strand,coding) expected in map-file > ?# passed through intstrand <- 3 ! > - > + > ? ? ? ?posf <- c("premakeped","mach") > ? ? ? ?intfmt <- which(posf == format)-1 > ? ? ? ?if (length(intfmt)!=1) {cat("format argument must be one of",posf,"\n");stop();} > - > + > ?# intfmt = 0 -> premakeped, merlin > ?# else -> mach > - > + > ? ? ? ?if (wslash) { > - ? ? ? ? ? ? ? .C("convert_snp_merlin_wslash",as.character(pedfile),as.character(mapfile),as.character(outfile),as.integer(intstrand),as.integer(bcast),as.character(alcodes),as.integer(length(alcodes)),as.integer(intfmt),as.integer(traits),PACKAGE="GenABEL") > + ? ? ? ? ? ? ? .C("convert_snp_merlin_wslash",as.character(pedfile),as.character(mapfile), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.character(outfile),as.integer(intstrand),as.integer(bcast), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.character(alcodes),as.integer(length(alcodes)),as.integer(intfmt), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.integer(traits),as.integer(mapHasHeaderLine), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? PACKAGE="GenABEL") > ? ? ? ?} else { > - ? ? ? ? ? ? ? .C("convert_snp_merlin",as.character(pedfile),as.character(mapfile),as.character(outfile),as.integer(intstrand),as.integer(bcast),as.character(alcodes),as.integer(length(alcodes)),as.integer(intfmt),as.integer(traits),PACKAGE="GenABEL") > + ? ? ? ? ? ? ? .C("convert_snp_merlin",as.character(pedfile),as.character(mapfile), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.character(outfile),as.integer(intstrand),as.integer(bcast), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.character(alcodes),as.integer(length(alcodes)),as.integer(intfmt), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.integer(traits),as.integer(mapHasHeaderLine), > + ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? PACKAGE="GenABEL") > ? ? ? ?} > - ? ? ? return(invisible(0)) > + ? ? ? return(invisible(0)) > ?} > > > Modified: pkg/GenABEL/man/convert.snp.ped.Rd > =================================================================== > --- pkg/GenABEL/man/convert.snp.ped.Rd ?2011-03-04 18:44:15 UTC (rev 666) > +++ pkg/GenABEL/man/convert.snp.ped.Rd ?2011-03-06 19:11:17 UTC (rev 667) > @@ -6,14 +6,15 @@ > ?} > ?\usage{ > ?convert.snp.ped(pedfile, mapfile, outfile, format = "premakeped", traits = 1, > - ? ? ? strand = "u", bcast = 10000000, wslash=F) > + ? ? ? strand = "u", bcast = 10000000, wslash=FALSE, mapHasHeaderLine=TRUE) > ?} > ?\arguments{ > ? \item{pedfile}{ > - ? ? ? Pre-makeped linkage genotypic data file name > + ? ? ? Pre-makeped linkage genotypic data file name (no header line) > ? ? ? ?} > ? \item{mapfile}{ > - ? ? ? Name of the map file > + ? ? ? Name of the map file (note that by default header line should be included, > + ? ? ? see 'mapHasHeaderLine' to modify that) > ? ? ? ?} > ? \item{outfile}{ > ? ? ? ?Output data file > @@ -41,6 +42,9 @@ > ? ? ? ?wslash=T it is assumed that genotypes are coded with single characters, > ? ? ? ?separated with slash (no spaces), e.g. "A/G", and not "A/ G" or "A / G". > ? ? ? ?} > + ?\item{mapHasHeaderLine}{ > + ? ? ? Whether map-file has a header line > + ? ? ? } > ?} > ?\details{ > ? ? ? ?Pedfile must be standard pre-makeped/Merlin linkage file, or a Mach file. > @@ -89,8 +93,9 @@ > > ? ? ? ?Accepted strand coding: +, -, u (unknown) > > - ? ? ? Please note that the header line (e.g. "chrom ? name ? ?position") SHOULD > - ? ? ? be present in your file! > + ? ? ? Please note that by default the header line (e.g. "chrom ? ? ? ?name ? ?position") SHOULD > + ? ? ? be present in your file, though you can use 'mapHasHeaderLine' argument to modify > + ? ? ? this behavior > > ?} > ?\value{ > > Modified: pkg/GenABEL/src/GAlib/convert_snp_merlin.cpp > =================================================================== > --- pkg/GenABEL/src/GAlib/convert_snp_merlin.cpp ? ? ? ?2011-03-04 18:44:15 UTC (rev 666) > +++ pkg/GenABEL/src/GAlib/convert_snp_merlin.cpp ? ? ? ?2011-03-06 19:11:17 UTC (rev 667) > @@ -41,7 +41,9 @@ > ?**/ > > ?extern "C" { > - ?void convert_snp_merlin (char** pedfilename, char** mapfilename, char** outfilename, int* Strandid, int* bcast, char **allele_codes, int* Ncodes, int *Fmt, int *Tra) { > + ?void convert_snp_merlin (char** pedfilename, char** mapfilename, char** outfilename, > + ? ? ? ? ? ? ? ? int* Strandid, int* bcast, char **allele_codes, int* Ncodes, > + ? ? ? ? ? ? ? ? int *Fmt, int *Tra, int *MapHasHeaderLine) { > > ? ? int verbose = *bcast ? 1 : 0; > > @@ -49,6 +51,7 @@ > ? ? int strandid = *Strandid; > ? ? int format = *Fmt; > ? ? int traits = *Tra; > + ? ?int mapHasHeaderLine = *MapHasHeaderLine; > > ? ? long int linecount=0; > ? ? string data; > @@ -83,7 +86,8 @@ > ? ? } > > ? ? int mapline=0; > - ? ?getline(mapfile,data); > +// read and ignore header line > + ? ?if (mapHasHeaderLine) getline(mapfile,data); > ? ? if (strandid==3) { > ? ? while (getline(mapfile,data)) { > ? ? ? istringstream datas (data); > > Modified: pkg/GenABEL/src/GAlib/convert_snp_merlin_wslash.cpp > =================================================================== > --- pkg/GenABEL/src/GAlib/convert_snp_merlin_wslash.cpp 2011-03-04 18:44:15 UTC (rev 666) > +++ pkg/GenABEL/src/GAlib/convert_snp_merlin_wslash.cpp 2011-03-06 19:11:17 UTC (rev 667) > @@ -40,7 +40,9 @@ > ?**/ > > ?extern "C" { > - ?void convert_snp_merlin_wslash (char** pedfilename, char** mapfilename, char** outfilename, int* Strandid, int* bcast, char **allele_codes, int* Ncodes, int *Fmt, int *Tra) { > + ?void convert_snp_merlin_wslash (char** pedfilename, char** mapfilename, > + ? ? ? ? ? ? ? ? char** outfilename, int* Strandid, int* bcast, char **allele_codes, > + ? ? ? ? ? ? ? ? int* Ncodes, int *Fmt, int *Tra, int *MapHasHeaderLine) { > > ? ? int verbose = *bcast ? 1 : 0; > > @@ -48,6 +50,7 @@ > ? ? int strandid = *Strandid; > ? ? int format = *Fmt; > ? ? int traits = *Tra; > + ? ?int mapHasHeaderLine = *MapHasHeaderLine; > > ? ? long int linecount=0; > ? ? string data; > @@ -82,7 +85,8 @@ > ? ? } > > ? ? int mapline=0; > - ? ?getline(mapfile,data); > +// read and ignore header line > + ? ?if (mapHasHeaderLine) getline(mapfile,data); > ? ? if (strandid==3) { > ? ? while (getline(mapfile,data)) { > ? ? ? istringstream datas (data); > > _______________________________________________ > Genabel-commits mailing list > Genabel-commits at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits > From l.karssen at erasmusmc.nl Mon Mar 7 14:34:50 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Mon, 07 Mar 2011 14:34:50 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References: Message-ID: <1299504890.1957.196.camel@rubidium> Hi all, One of the things that I think needs to be done before "public release" is to change some DNS settings. This would take 10-15 minutes. Right now only forum.genabel.org is working as expected. For the website (Drupal) we have www.genabel.org planned, right? I think that for consistency we should also set the correct DNS entries for genabel.com and genabel.net. Furthermore, do we want to point devel.genabel.* and/or dev.genabel.* to https://r-forge.r-project.org/projects/genabel/ ? And how about man.genabel.* -> http://test.genabel.org/drupal6/?q=manuals doc.genabel.* -> idem dito Any other suggestions? I added some notes below as well. On Fri, 2011-02-11 at 20:44 +0700, Yurii Aulchenko wrote: > I am sending a short description of web-server todo's to keep things documented > > Here are the things we need to do about web-server (add more and join > into discussion :) ) > > * Solve links (will tell more about that below) Also see below. We should also check the links for absolute references to http://test.genabel.org/drupal6/etc. I suggest to make all links relative: link. To automate this we could try to think of a search/replace action on the Drupal database (after a backup of course :-)). Furthermore, we can configure the web server to "strip" the /drupal6/ part as well. The URL of the pages will then be like this: www.genabel.org/?q=acknowledgements > > * Solve HTML documentation (will tell more about that below) > > * Ask PredictA and ParallA developers to generate HTML -- this will be > handy if they include this bit as a part of their package update > process (Yurii) We will also need to write instructions on how to do > that and put @developers' pages (other primary link at > www.genabel.org? R-forge?) > > * Check integrity -- all links work, and bring to right place I came across a Firefox extension that can validate the links on a page. It gives a visual indication (green/red links) for each link on a page which shows if the links points to an existing page. What it doesn't do, of course, is check whether the link goes to the intended place. That requires human intelligence, I'd say. However it doesn't give some sort of aggregated output, you'll have to visit each page and run the link checker. On the other hand, it beats going over the HTML code by hand :-). The URL of the extension is http://www.kevinfreitas.net/extensions/linkchecker/ > * Replace MGA *ABEL pages with something, which re-directs the traffic > to www.genabel.org This depends a bit on the level of access that you have on MGA. IIRC you have the files in your ~/public_html dir, right? There are a couple of ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite enabled we can use that. See e.g. http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an example. Otherwise the .htaccess file can be used. See e.g. http://www.stevenhargrove.com/redirect-web-pages/#301htaccess for an example. Steven Hargrove's page lists a few more examples. Lennart. > > * Announce (forum, mail-list, social networks) > > Now some more details about 'problems' > > (*) Solve links > > Primary links look like http://test.genabel.org/drupal6/?q=node/6 > (this is to Packages section). Can we have something more > human-readable? I believe Lennart figured what module we need to > enable, but doing so was not so trivial... > > (*) Solve HTML documentation > > We need to keep HTML documentation for packages, and eventually > update, keeping the link stable. > > This is a bad one, though there is workaround. What we have figured > out with Lennart about htmlimport module, is that if you import html > pages (e.g. from dir http://test.genabel.org/old/htmlMetA/), the > delete it in drupal, and then try to import again, it will not work. > So now trying to import http://test.genabel.org/old/htmlMetA/ > generates crap, and that's it. The workaround is to say rename > htmlMetA to say htmlMetA1, then import, -- it will work. But! It will > bring htmlMetA1 into the link's name, so the link to 'updated' version > changes! > > So, for the moment I just put all docs to http://test.genabel.org/old/ > and make links to there. It is kind of ok, but two problems here I > believe: (a) search engines will eventually index 'old', and people > will land @ old without noticing we have this beautiful web-site :) > (b) just esthetically would prefer to see docs appearing "within" > drupal frame, see example > > in old: http://test.genabel.org/old/htmlDatA/00Index.html > > how it should be: > http://test.genabel.org/drupal6/?q=htmlDatA/00Index, and it should be > linked from http://test.genabel.org/drupal6/?q=node/84 > > I wonder -- if you just try to import updated page, without prior > delete of pages to be updated, would that work just fine solving the > problem? > > If nothing works, would that be reason good enough to try other CMS or > even go to very simple plain HTML site?... > > best wishes, > Yurii > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 email: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From pirastu at burlo.trieste.it Mon Mar 7 14:36:13 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Mon, 7 Mar 2011 14:36:13 +0100 (CET) Subject: [GenABEL-dev] descriptives.traits by.var bugs. Message-ID: <53018.212.189.217.13.1299504973.squirrel@webmail-rupar.regione.fvg.it> Dear all, I'm new to the group so I'm not sure I'm doing this correctly, so if I'm not please tell me so. Anyways I think I found how to fix the problems with descriptives.traits: 1) descriptives.trait with missings in by.var argument this line is causing the problem if (!is.numeric(ctrao) | all(ctrao == svar)) it should be like this: if (!is.numeric(ctrao) | all(ctrao == svar,na.rm=T)) 2) descriptives.trait can not see the variable in by.var argument unless full path specified. the line svar <- by.var and could be modified with these 3 lines: if(is.character(by.var) & length(by.var)==1){ svar <- data[,by.var]} #in this case by.var should be specified as ie. "sex" if(is.vector(by.var)){ svar <- by.var }# in this case by.var should be like phdata(ge03d2ex)$sex if(is.formula(by.var)){ svar <- data[,as.character(by.var)[2]]} #in this case by.var should have the form ie. ~sex I hope I could be of help. Sorry if I did not do things the right way. Best Nicola From yurii.aulchenko at gmail.com Mon Mar 7 14:52:32 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 14:52:32 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: <1299504890.1957.196.camel@rubidium> References: <1299504890.1957.196.camel@rubidium> Message-ID: > One of the things that I think needs to be done before "public release" > is to change some DNS settings. This would take 10-15 minutes. > Right now only forum.genabel.org is working as expected. For the website > (Drupal) we have www.genabel.org planned, right? > I think that for consistency we should also set the correct DNS entries > for genabel.com and genabel.net. > > Furthermore, do we want to point devel.genabel.* and/or dev.genabel.* to > https://r-forge.r-project.org/projects/genabel/ ? > And how about > man.genabel.* -> http://test.genabel.org/drupal6/?q=manuals > doc.genabel.* -> idem dito > These are terrific suggestions, thanks, Lennart. Cause we need to plan the time, one of the main pieces I see in your message that migrating from test.genabel.org to www.genabel.org will take 15 minute only. These are good news :) >> * Solve links (will tell more about that below) > > Also see below. > > We should also check the links for absolute references to > http://test.genabel.org/drupal6/etc. I suggest to make all links > relative: link. > To automate this we could try to think of a search/replace action on the > Drupal database (after a backup of course :-)). > Not sure I quite follow, I think you and Anatoly have better idea and leave it up to you guys to decide :) > Furthermore, we can configure the web server to "strip" the /drupal6/ > part as well. The URL of the pages will then be like this: > www.genabel.org/?q=acknowledgements This sounds like a good idea! >> * Check integrity -- all links work, and bring to right place > > I came across a Firefox extension that can validate the links on a page. > It gives a visual indication (green/red links) for each link on a page > which shows if the links points to an existing page. What it doesn't do, > of course, is check whether the link goes to the intended place. That > requires human intelligence, I'd say. However it doesn't give some sort > of aggregated output, you'll have to visit each page and run the link > checker. On the other hand, it beats going over the HTML code by > hand :-). > The URL of the extension is > http://www.kevinfreitas.net/extensions/linkchecker/ > This should come handy, thanks! >> * Replace MGA *ABEL pages with something, which re-directs the traffic >> to www.genabel.org > > This depends a bit on the level of access that you have on MGA. IIRC you > have the files in your ~/public_html dir, right? There are a couple of > ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite > enabled we can use that. See e.g. > http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an > example. > Wow. I thought of something more simple, which I used for old-old-old page of GenABEL, see http://mga.bionet.nsc.ru/nlru/GenABEL basically, it boils down to this HTML line Anyone sees any problem to use this? best wishes, Yurii From l.karssen at erasmusmc.nl Mon Mar 7 15:01:49 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Mon, 07 Mar 2011 15:01:49 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References: <1299504890.1957.196.camel@rubidium> Message-ID: <1299506509.1957.206.camel@rubidium> On Mon, 2011-03-07 at 14:52 +0100, Yurii Aulchenko wrote: > > One of the things that I think needs to be done before "public release" > > is to change some DNS settings. This would take 10-15 minutes. > > Right now only forum.genabel.org is working as expected. For the website > > (Drupal) we have www.genabel.org planned, right? > > I think that for consistency we should also set the correct DNS entries > > for genabel.com and genabel.net. > > > > Furthermore, do we want to point devel.genabel.* and/or dev.genabel.* to > > https://r-forge.r-project.org/projects/genabel/ ? > > And how about > > man.genabel.* -> http://test.genabel.org/drupal6/?q=manuals > > doc.genabel.* -> idem dito > > > > These are terrific suggestions, thanks, Lennart. Cause we need to plan > the time, one of the main pieces I see in your message that migrating > from test.genabel.org to www.genabel.org will take 15 minute only. > These are good news :) Even though it doesn't take much time I think we should do this early because it will help with the link stuff further down the road. This way any errors in the links (e.g. those with absolute paths like http://test.genabel.org/etc) will show up early and allow us more time for fixing. > >> * Replace MGA *ABEL pages with something, which re-directs the traffic > >> to www.genabel.org > > > > This depends a bit on the level of access that you have on MGA. IIRC you > > have the files in your ~/public_html dir, right? There are a couple of > > ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite > > enabled we can use that. See e.g. > > http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an > > example. > > > > Wow. I thought of something more simple, which I used for old-old-old > page of GenABEL, see > > http://mga.bionet.nsc.ru/nlru/GenABEL > > basically, it boils down to this HTML line > > > > Anyone sees any problem to use this? It really is "the old way" of doing it and it may incur search engine penalties. Check the "HTML Redirection" section at the top of on Steven Hargrove's web site for some arguments against it. However, it will work as expected. Regards, Lennart. > > best wishes, > Yurii -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 email: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From yurii.aulchenko at gmail.com Mon Mar 7 15:16:36 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 15:16:36 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: <1299506509.1957.206.camel@rubidium> References: <1299504890.1957.196.camel@rubidium> <1299506509.1957.206.camel@rubidium> Message-ID: > Even though it doesn't take much time I think we should do this early > because it will help with the link stuff further down the road. This way > any errors in the links (e.g. those with absolute paths like > http://test.genabel.org/etc) will show up early and allow us more time > for fixing. > Ok, point taken. We have to have our site up, running, and in good shape by March 14 eve latest. A number of people are mentioned on 'People' page, and we need to let them see the site before it comes public, and we can also ask them to check links etc.. We need to allow few days for the feedback. So, I suggest that we plan to * have site-ready-for-checks on March 9 evening * migrate to other DNS March 10 * ask people to have a look on March 10 evening / March 11 morning > It really is "the old way" of doing it and it may incur search engine > penalties. Check the "HTML Redirection" section at the top of on Steven > Hargrove's web site for some arguments against it. > However, it will work as expected. Clear. I think the old way should do between March 10/11 and March 15, and we should take better care of this after March 15. best wishes, Yurii From xia.shen at lcb.uu.se Mon Mar 7 16:52:00 2011 From: xia.shen at lcb.uu.se (Xia Shen) Date: Mon, 7 Mar 2011 16:52:00 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se>

Message-ID: <3CF9915B-8747-405B-B222-846EEF3F9471@lcb.uu.se> Hi Yurii, I just 'sudo-randomly' tried some different argument value in the hglm function call, and happened to have solved the bias issue (at least it seems to be so). Please try putting the following arguments in your hglm wrapper: hglm(y = ph, X = desmat, Z = L, conv = 1e-6, maxit = 40) I re-did 100 simulations and the bias is not sig from 0 with p-value = 0.77! I guess the problem here is just numerical. Since IBS matrix is generally sparse (close to identity), the estimation of variance components will be sensitive numerically. Just make the convergence criteria more strict, and it works. Note that this might slow down hglm a bit (not very much I guess), but I think it should be more than 20x faster than the old polygenic. Not familiar with fmm, but: 1. our hglm can deal with any GLM family of the trait distribution, even non-normal genetic effects. 2. we produce standard errors for everything - every part is checkable using h-likelihood :) Cheers Xia On Feb 28, 2011, at 12:24 AM, Yurii Aulchenko wrote: > some extra info: > > was running the checks cf. simulated heritability > > t-test > > difference (h2simulated - h2estimated-old-polygenic) ns from 0, p=0.36 > > difference (h2simulated - h2estimated-hglm) SIG from 0, p=0.006 > > difference (h2simulated - h2estimated-fmm) ns from 0, p=0.30 > > So, my feeling is that hglm does something wrong; old and fmm do very > well, but old does better :) > > On Sun, Feb 27, 2011 at 10:51 PM, Yurii Aulchenko > wrote: >> Dear All, >> >> Some of you may know, that MixABEL::FastMixedModel by William Astle >> first does polygenic model estimation (e.g. h2-estimate is given by >> FastMixedModel(...)$null.herit), though it is build 'within' GWA and >> it is not yet possible to get hold on all necessary 'null' >> heritability estimates. >> > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel From yurii.aulchenko at gmail.com Mon Mar 7 17:50:13 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 17:50:13 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: <3CF9915B-8747-405B-B222-846EEF3F9471@lcb.uu.se> References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se>

<3CF9915B-8747-405B-B222-846EEF3F9471@lcb.uu.se> Message-ID: Thanks, will update the code, give it a try in larger # of simulations, and tell the results! Yurii On Mon, Mar 7, 2011 at 4:52 PM, Xia Shen wrote: > Hi Yurii, > > I just 'sudo-randomly' tried some different argument value in the hglm function call, and happened to have solved the bias issue (at least it seems to be so). Please try putting the following arguments in your hglm wrapper: > > hglm(y = ph, X = desmat, Z = L, conv = 1e-6, maxit = 40) > > I re-did 100 simulations and the bias is not sig from 0 with p-value = 0.77! > > I guess the problem here is just numerical. Since IBS matrix is generally sparse (close to identity), the estimation of variance components will be sensitive numerically. Just make the convergence criteria more strict, and it works. > > Note that this might slow down hglm a bit (not very much I guess), but I think it should be more than 20x faster than the old polygenic. Not familiar with fmm, but: 1. our hglm can deal with any GLM family of the trait distribution, even non-normal genetic effects. 2. we produce standard errors for everything - every part is checkable using h-likelihood :) > > Cheers > > Xia > From yurii.aulchenko at gmail.com Mon Mar 7 18:14:24 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 18:14:24 +0100 Subject: [GenABEL-dev] [Genabel-commits] r684 - in pkg/VariABEL: . R man In-Reply-To: <20110307171010.5F4999CA83@r-forge.r-project.org> References: <20110307171010.5F4999CA83@r-forge.r-project.org> Message-ID: This commit comment suggest this version should pass R CMD check ... From yurii.aulchenko at gmail.com Mon Mar 7 18:17:39 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 18:17:39 +0100 Subject: [GenABEL-dev] Fwd: [Genabel-commits] r684 - in pkg/VariABEL: . R man In-Reply-To: References: <20110307171010.5F4999CA83@r-forge.r-project.org> Message-ID: Sorry, somehow my e-mail got cut at the start, once again This commit comment suggest this version should pass R CMD check From R CMD INSTALL I get ** building package indices ... ** testing if installed package can be loaded Error in namespaceExport(ns, exports) : undefined exports: var.test.gwaa ERROR: loading failed * removing ?/Library/Frameworks/R.framework/Resources/library/VariABEL? * restoring previous ?/Library/Frameworks/R.framework/Resources/library/VariABEL? It looks like in NAMESPACE you export var.test.gwaa, while the function name is var_test_gwaa Yurii On Mon, Mar 7, 2011 at 6:10 PM, ? wrote: > Author: maksim > Date: 2011-03-07 18:10:10 +0100 (Mon, 07 Mar 2011) > New Revision: 684 > > Added: > ? pkg/VariABEL/man/var_test_gwaa.Rd > Removed: > ? pkg/VariABEL/man/var.test.homogeneity.Rd > Modified: > ? pkg/VariABEL/NAMESPACE > ? pkg/VariABEL/R/var.meta.gwaa.R > ? pkg/VariABEL/R/var.test.gwaa.R > Log: > fix errors from R CMD check > > Modified: pkg/VariABEL/NAMESPACE > =================================================================== > --- pkg/VariABEL/NAMESPACE ? ? ?2011-03-07 15:18:02 UTC (rev 683) > +++ pkg/VariABEL/NAMESPACE ? ? ?2011-03-07 17:10:10 UTC (rev 684) > @@ -2,7 +2,7 @@ > > > ?export( > - ? ? ? var.meta.gwaa, > +# ? ? ?var.meta.gwaa, > ? ? ? ?var.test.gwaa#, > ? ? ? ?#test_databel > ? ? ? ?) > > Modified: pkg/VariABEL/R/var.meta.gwaa.R > =================================================================== > --- pkg/VariABEL/R/var.meta.gwaa.R ? ? ?2011-03-07 15:18:02 UTC (rev 683) > +++ pkg/VariABEL/R/var.meta.gwaa.R ? ? ?2011-03-07 17:10:10 UTC (rev 684) > @@ -16,36 +16,36 @@ > ?#===================================================================================== > > > - > - > -"var.meta.gwaa" <- > -function(input_filenames, > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?testname, > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?output_filename="output.variance.metaanalysis", > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?exclude_snp_below_threshold=F, > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?threshold=30, > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?all_warnings=F, > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?skip_first_lines_amount=0, > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?delim=' ') { > - > - > -file_amount <- length(filenames_array) > - > - > -return_val <- ?.C("var_meta_gwaa_C", input_filenames, as.integer(file_amount), output_filename, as.integer(skip_first_lines_amount), delim, > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? lambdas = double(file_amount+1), > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? lambdas_NA = integer(file_amount+1), > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.logical(exclude_snp_below_threshold), > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.integer(threshold), > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.logical(all_warnings), > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? as.character(testname)) From yurii.aulchenko at gmail.com Mon Mar 7 19:24:32 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 19:24:32 +0100 Subject: [GenABEL-dev] descriptives.traits by.var bugs. In-Reply-To: <53018.212.189.217.13.1299504973.squirrel@webmail-rupar.regione.fvg.it> References: <53018.212.189.217.13.1299504973.squirrel@webmail-rupar.regione.fvg.it> Message-ID: Nicola, many thanks for this! Questions: we have few 'by.var' args bugs filed in at https://r-forge.r-project.org/search/?type_of_search=full&group_id=505&atid=0&words=by.var&Search=Search I think your suggestions take direct care of [#1185] descriptives.trait with missings in by.var argument [#1184] descriptives.trait can not see the variable in by.var argument unless full path specified correct? did you try to see if it would address [#1259] descriptives.trait with by.var argument does not work (contributed by Richard Ahn) as well? best wishes, and thanks a lot! Yurii On Mon, Mar 7, 2011 at 2:36 PM, wrote: > Dear all, > > I'm new to the group so I'm not sure I'm doing this correctly, so if I'm > not please tell me so. > Anyways I think I found how to fix the problems with descriptives.traits: > > 1) descriptives.trait with missings in by.var argument > > ? this line is causing the problem > > ? if (!is.numeric(ctrao) | all(ctrao == svar)) > > it should be like this: > > if (!is.numeric(ctrao) | all(ctrao == svar,na.rm=T)) > > > 2) ?descriptives.trait can not see the variable in by.var argument unless > full path specified. > > the line > svar <- by.var > > and could be modified with these 3 lines: > > if(is.character(by.var) & length(by.var)==1){ svar <- data[,by.var]} #in > this case by.var should be specified as ie. "sex" > > if(is.vector(by.var)){ svar <- by.var }# in this case by.var should be > like phdata(ge03d2ex)$sex > > if(is.formula(by.var)){ svar <- data[,as.character(by.var)[2]]} #in this > case by.var should have the form ie. ~sex > > > I hope I could be of help. Sorry if I did not do things the right way. > > Best > > Nicola > > > > > > > > > > > > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > From pirastu at burlo.trieste.it Mon Mar 7 21:21:04 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Mon, 7 Mar 2011 21:21:04 +0100 (CET) Subject: [GenABEL-dev] descriptives.traits by.var bugs. In-Reply-To: References: <53018.212.189.217.13.1299504973.squirrel@webmail-rupar.regione.fvg.it> Message-ID: <49354.93.48.255.18.1299529264.squirrel@webmail-rupar.regione.fvg.it> I've just looked at the problem [#1259] descriptives.trait with by.var argument does not work here's the problem he's had : descriptives.trait(data=test, by = bt) and it's the same you mentioned about sex, so actually it does fix it. Best Nicola > Nicola, > > many thanks for this! > > Questions: we have few 'by.var' args bugs filed in at > > https://r-forge.r-project.org/search/?type_of_search=full&group_id=505&atid=0&words=by.var&Search=Search > > I think your suggestions take direct care of > > [#1185] descriptives.trait with missings in by.var argument > [#1184] descriptives.trait can not see the variable in by.var argument > unless full path specified > > correct? > > did you try to see if it would address > > [#1259] descriptives.trait with by.var argument does not work > (contributed by Richard Ahn) > > as well? > > best wishes, and thanks a lot! > > Yurii > > > On Mon, Mar 7, 2011 at 2:36 PM, wrote: >> Dear all, >> >> I'm new to the group so I'm not sure I'm doing this correctly, so if I'm >> not please tell me so. >> Anyways I think I found how to fix the problems with >> descriptives.traits: >> >> 1) descriptives.trait with missings in by.var argument >> >> ? this line is causing the problem >> >> ? if (!is.numeric(ctrao) | all(ctrao == svar)) >> >> it should be like this: >> >> if (!is.numeric(ctrao) | all(ctrao == svar,na.rm=T)) >> >> >> 2) ?descriptives.trait can not see the variable in by.var argument >> unless >> full path specified. >> >> the line >> svar <- by.var >> >> and could be modified with these 3 lines: >> >> if(is.character(by.var) & length(by.var)==1){ svar <- data[,by.var]} #in >> this case by.var should be specified as ie. "sex" >> >> if(is.vector(by.var)){ svar <- by.var }# in this case by.var should be >> like phdata(ge03d2ex)$sex >> >> if(is.formula(by.var)){ svar <- data[,as.character(by.var)[2]]} #in this >> case by.var should have the form ie. ~sex >> >> >> I hope I could be of help. Sorry if I did not do things the right way. >> >> Best >> >> Nicola >> >> >> >> >> >> >> >> >> >> >> >> >> _______________________________________________ >> genabel-devel mailing list >> genabel-devel at lists.r-forge.r-project.org >> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel >> > From yurii.aulchenko at gmail.com Mon Mar 7 22:10:45 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 22:10:45 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: <3CF9915B-8747-405B-B222-846EEF3F9471@lcb.uu.se> References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se>

<3CF9915B-8747-405B-B222-846EEF3F9471@lcb.uu.se> Message-ID: Xia et al., here are results > I just 'sudo-randomly' tried some different argument value in the hglm function call, and happened to have solved the bias issue (at least it seems to be so). Please try putting the following arguments in your hglm wrapper: > > hglm(y = ph, X = desmat, Z = L, conv = 1e-6, maxit = 40) > > I re-did 100 simulations and the bias is not sig from 0 with p-value = 0.77! > To be even more strict, I have set hglm( ... , conv = 1e-8, maxit = 100) and ran the tests. Results attached as PDF (+ the script) First, some great news: See figure, columns a) and b): there is almost 1-to-1 identity in h2 estimates generated by polygenic_hglm and FMM. I take this as a sign that both of them do a great job, while 'old' polygenic underestimates h2 slightly. Do not quite understand whathappens with old -- and not quite willing to dig into it yet, as lambdas are closest to 1 with old -- so I think we stick to the old form of old-polygenic till we can explain the whole story. Will wait till polygenic-FMM is out to do another round of comparisons. So, I thin Xia's suggestion > I guess the problem here is just numerical. Since IBS matrix is generally sparse (close to identity), the estimation of variance components will be sensitive numerically. Just make the convergence criteria more strict, and it works. > is right. And the 0.4-strangeness must be something to deal with the starting values, what do you think? > Note that this might slow down hglm a bit (not very much I guess), but I think it should be more than 20x faster than the old polygenic. Not familiar with fmm, but: 1. our hglm can deal with any GLM family of the trait distribution, even non-normal genetic effects. 2. we produce standard errors for everything - every part is checkable using h-likelihood :) > It did become slower with the strict criteria -- now speed-up is x5.3 (range: x1.7 - x80.8). Still, this is quite incredible :) Next step will be William's polygenic_fmm -- his speed-up is more-than-incredible :) Ok, what about finalizing this so we have 'polygenic_hglm' ready-to-use in the next release? Here is a plan: I will take care of updating documentation to 'polygenic' so it references to 'polygenic_hglm' as an alternative; also main package documentation. You will see that on genabel-commits. ---- Things we absolutely need before it can be released: * Test if resulting object works at all with grammar procedure and produces roughly similar results with old-polygenic object * Pull the se's of the estimates in the output -- this allows for testing significance of covariates * Update documentation: ** Xia, please make sure you mention everything you want to mention -- give proper credit, references, etc. ** Better explanation of the method. We can of cause always refer people to the main 'hglm' man page, but it will be nice to outline -- e.g. that REML is used by default ** Examples: how do you test for a SNP effect? See for example of situation with old polygenic: http://forum.genabel.org/viewtopic.php?f=6&t=134&sid=c5dcf7fc39fd3f7d07b1308bafed5286 ----- Things, which will be great to have, but not critical to the release: ** Similar to old-polygenic fixh2 argument -- in which case we can also output -2xLogLik (but need to make sure people understand that eg in REML setting this can only be used for testing random effect / h2) [also look up what are other options available in old-polygenic but not in polygenic_hglm] ** optimization of the code -- many things which are computed in wrapper should have been computed by 'hglm', so we can gain some speed ---- Nice, but not-so-urgent things: * unify and formalize 'polygenic' as S4-class best wishes, and hear from you, Yurii PS To be followed-up later: why the hell does old-polygenic+mmscore still generates Lambdas closest to 1 (see figure, columns c), d))? Something to deal with uncertainty in gkin, as suggested by William (very interesting question!), conservativity of the 'mmscore' as suggested by Yurii, etc. At the end, we want to know which one is the most powerful procedure. This is something we can test directly. -------------- next part -------------- A non-text attachment was scrubbed... Name: test_old_hglm_FMM.pdf Type: application/pdf Size: 149460 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: test_polygenic_hglm_fmm_2011.03.07.R Type: application/octet-stream Size: 3723 bytes Desc: not available URL: From yurii.aulchenko at gmail.com Mon Mar 7 22:51:07 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 22:51:07 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References: <1299504890.1957.196.camel@rubidium> <1299506509.1957.206.camel@rubidium> Message-ID: In the next couple of days I am going to mess around with our web-pages (prototype at http://test.genabel.org/drupal6/). If anyone is willing to help (but mind strict deadlines) please propose what specific actions you can help with. General suggestions + proof-reading pages are most welcome! What I propose to do is 1) re-structure 'Packages' -- separate sub-page for every package 2) generate HTML manuals for ParallABEL and PredictABEL and put these to http://test.genabel.org/old/ 3) go over all pages and check the content -- updates, awkward phrasing, etc. 4) have a go at the devel-pages @RForge. I will put our 'how to install devel-version' there, to be followed up with other devel documents 5) introduce 'For developers' primary link, pointing to RForge 6) remove 'DatABEL manual index' from Tutorials -- this should clearly go to manuals I will generate HTML on SVN and put to test.genabel...-drupal immediately. My work will have to be followed-up: (2) -> Anatoly, as soon this will be ready, could you please be prepared to import all manuals html (7, at http://test.genabel.org/old/) + make proper links from the 'manuals' pages. We need references to all 7 manuals there. Also, create secondary nodes for the manuals -- as we have with packages and tutorials. * Links: for many links I will provide the description [ here-a-link-should-point-to... ] , but not the links. Will need help of Anatoly to take real-word treatment of this Please reply if you see there is a better structure/plan, something extra should be taken care of, any suggestions etc. Anybody interested, please follow my actions at SVN and genabel-commits list, and reply with your suggestions to this list. If you can read HTML, your help is really needed! :) best wishes, Yurii From yurii.aulchenko at gmail.com Mon Mar 7 23:47:42 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 23:47:42 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References: <1299504890.1957.196.camel@rubidium> <1299506509.1957.206.camel@rubidium>

Message-ID: > 1) re-structure 'Packages' -- separate sub-page for every package done! Anatoly/Lennart -- can you please have a look if I filled in links all right (I was trying to make relative references). Let me know if I should modify my behavior! > 2) generate HTML manuals for ParallABEL and PredictABEL and put these > to http://test.genabel.org/old/ done! Now, Anatoly -- this is your move. We need all this htmls imported and listed under the 'manuals' node That's it for today. I have a feeling I am talking a bit to myself -- which I do :P best wishes, and good night, Yurii From yurii.aulchenko at gmail.com Mon Mar 7 23:57:21 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 7 Mar 2011 23:57:21 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References: Message-ID: Just an idea: can we connect our web-site news feed to the news feed of the forum.genabel.org? If not, just need to make sure they are in sync. From pirastu at burlo.trieste.it Tue Mar 8 00:22:01 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Tue, 8 Mar 2011 00:22:01 +0100 (CET) Subject: [GenABEL-dev] descriptives.traits by.var bugs. In-Reply-To: <49354.93.48.255.18.1299529264.squirrel@webmail-rupar.regione.fvg.it> References: <53018.212.189.217.13.1299504973.squirrel@webmail-rupar.regione.fvg.it> <49354.93.48.255.18.1299529264.squirrel@webmail-rupar.regione.fvg.it> Message-ID: <49938.93.48.255.18.1299540121.squirrel@webmail-rupar.regione.fvg.it> Sorry, I took a better look and the problem and it is sligthly different, in fact it's not the script but the dataset. Basically all of the controls are missing the variable age and the t.test crushes and gives out that error. I think a check should be added to avoid this from happening, meanwhile I've posted a reply on the forum on how to get results anyways. Best Nicola > I've just looked at the problem > > [#1259] descriptives.trait with by.var argument does not work > > here's the problem he's had : descriptives.trait(data=test, by = bt) > > and it's the same you mentioned about sex, so actually it does fix it. > > Best > > Nicola > >> Nicola, >> >> many thanks for this! >> >> Questions: we have few 'by.var' args bugs filed in at >> >> https://r-forge.r-project.org/search/?type_of_search=full&group_id=505&atid=0&words=by.var&Search=Search >> >> I think your suggestions take direct care of >> >> [#1185] descriptives.trait with missings in by.var argument >> [#1184] descriptives.trait can not see the variable in by.var argument >> unless full path specified >> >> correct? >> >> did you try to see if it would address >> >> [#1259] descriptives.trait with by.var argument does not work >> (contributed by Richard Ahn) >> >> as well? >> >> best wishes, and thanks a lot! >> >> Yurii >> >> >> On Mon, Mar 7, 2011 at 2:36 PM, wrote: >>> Dear all, >>> >>> I'm new to the group so I'm not sure I'm doing this correctly, so if >>> I'm >>> not please tell me so. >>> Anyways I think I found how to fix the problems with >>> descriptives.traits: >>> >>> 1) descriptives.trait with missings in by.var argument >>> >>> ? this line is causing the problem >>> >>> ? if (!is.numeric(ctrao) | all(ctrao == svar)) >>> >>> it should be like this: >>> >>> if (!is.numeric(ctrao) | all(ctrao == svar,na.rm=T)) >>> >>> >>> 2) ?descriptives.trait can not see the variable in by.var argument >>> unless >>> full path specified. >>> >>> the line >>> svar <- by.var >>> >>> and could be modified with these 3 lines: >>> >>> if(is.character(by.var) & length(by.var)==1){ svar <- data[,by.var]} >>> #in >>> this case by.var should be specified as ie. "sex" >>> >>> if(is.vector(by.var)){ svar <- by.var }# in this case by.var should be >>> like phdata(ge03d2ex)$sex >>> >>> if(is.formula(by.var)){ svar <- data[,as.character(by.var)[2]]} #in >>> this >>> case by.var should have the form ie. ~sex >>> >>> >>> I hope I could be of help. Sorry if I did not do things the right way. >>> >>> Best >>> >>> Nicola >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> _______________________________________________ >>> genabel-devel mailing list >>> genabel-devel at lists.r-forge.r-project.org >>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel >>> >> > > > From yurii.aulchenko at gmail.com Tue Mar 8 00:26:55 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 8 Mar 2011 00:26:55 +0100 Subject: [GenABEL-dev] [Genabel-commits] r688 - pkg/GenABEL/R In-Reply-To: <20110307232601.450049CA83@r-forge.r-project.org> References: <20110307232601.450049CA83@r-forge.r-project.org> Message-ID: As promised, updated polygenic_hglm parameters + did the documentation cross-linking Yurii On Tue, Mar 8, 2011 at 12:26 AM, wrote: > Author: yurii > Date: 2011-03-08 00:26:01 +0100 (Tue, 08 Mar 2011) > New Revision: 688 > > Modified: > ? pkg/GenABEL/R/GenABEL-package.R > ? pkg/GenABEL/R/polygenic.R > ? pkg/GenABEL/R/polygenic_hglm.R > Log: > update on polygenic_hglm parameters + documentation cross-linking > > Modified: pkg/GenABEL/R/GenABEL-package.R > =================================================================== > --- pkg/GenABEL/R/GenABEL-package.R ? ? 2011-03-07 23:24:59 UTC (rev 687) > +++ pkg/GenABEL/R/GenABEL-package.R ? ? 2011-03-07 23:26:01 UTC (rev 688) > @@ -104,6 +104,7 @@ > ?#' \code{\link{ibs}} (computations of IBS / genomic IBD), > ?#' \code{\link{egscore}} (stratification adjustment following Price et al.), > ?#' \code{\link{polygenic}} (heritability analysis), > +#' \code{\link{polygenic_hglm}} (another function for heritability analysis), > ?#' \code{\link{mmscore}} (score test of Chen and Abecasis), > ?#' \code{\link{grammar}} (grammar test of Aulchenko et al.). > ?#' > > Modified: pkg/GenABEL/R/polygenic.R > =================================================================== > --- pkg/GenABEL/R/polygenic.R ? 2011-03-07 23:24:59 UTC (rev 687) > +++ pkg/GenABEL/R/polygenic.R ? 2011-03-07 23:26:01 UTC (rev 688) > @@ -129,7 +129,12 @@ > ?#' If a trait (no covarites) is used, make sure that order of IDs in > ?#' kinship.matrix is exactly the same as in the outcome > ?#' > +#' Please note that there is alternative to 'polygenic', > +#' \code{\link{polygenic_hglm}}, which is faster than > +#' 'polygenic'. > +#' > ?#' @seealso > +#' \code{\link{polygenic_hglm}}, > ?#' \code{\link{mmscore}}, > ?#' \code{\link{grammar}} > ?#' > > Modified: pkg/GenABEL/R/polygenic_hglm.R > =================================================================== > --- pkg/GenABEL/R/polygenic_hglm.R ? ? ?2011-03-07 23:24:59 UTC (rev 687) > +++ pkg/GenABEL/R/polygenic_hglm.R ? ? ?2011-03-07 23:26:01 UTC (rev 688) > @@ -2,6 +2,8 @@ > ?#' > ?#' Estimation of polygenic model using hierarchical > ?#' GLM (hglm package) > +#' Tell when REML is used, what otherwise; > +#' what it is doing, how to do tests, add more examples, references, etc. > ?#' > ?#' @param formula Formula describing fixed effects to be used in analysis, e.g. > ?#' y ~ a + b means that outcome (y) depends on two covariates, a and b. > @@ -14,6 +16,8 @@ > ?#' @param family a description of the error distribution and link function to be > ?#' used in the mean part of the model (see \code{\link{family}} for details of > ?#' family functions) > +#' @param conv 'conv' parameter of 'hglm' (stricter then default) > +#' @param maxit 'maxit' parameter of 'hglm' (stricter then default) > ?#' @param ... other parameters passed to 'hglm' call > ?#' > ?#' @author Xia Shen, Yurii Aulchenko > @@ -22,9 +26,9 @@ > ?#' need to put reference here > ?#' > ?#' @seealso > +#' \code{\link{polygenic}}, > ?#' \code{\link{mmscore}}, > ?#' \code{\link{grammar}} > -#' \code{\link{polygenic}} > ?#' > ?#' @examples > ?#' data(ge03d2ex.clean) > @@ -46,7 +50,7 @@ > ?#' > ?#' @keywords htest > ?#' > -"polygenic_hglm" <- function(formula,kinship.matrix,data,family=gaussian(), ... ) > +"polygenic_hglm" <- function(formula,kinship.matrix,data,family=gaussian(), conv=1e-8, maxit=100, ... ) > ?{ > ? ? ? ?if (!require(hglm)) > ? ? ? ? ? ? ? ?stop("this function requires 'hglm' package to be installed") > @@ -71,7 +75,9 @@ > ? ? ? ?relmat <- relmat*2.0 > ? ? ? ?s <- svd(relmat) > ? ? ? ?L <- s$u %*% diag(sqrt(s$d)) > - ? ? ? res_hglm <- hglm(y = y, X = desmat, Z = L, family = family, ... ) > + ? ? ? cnv <- 1e-6 > + ? ? ? mxit <- 40 > + ? ? ? res_hglm <- hglm(y = y, X = desmat, Z = L, family = family, conv = conv, maxit = maxit, ... ) > ? ? ? ?#sum_res_hglm <- summary(res_hglm) > > ? ? ? ?out <- list() > > _______________________________________________ > Genabel-commits mailing list > Genabel-commits at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-commits > From yurii.aulchenko at gmail.com Tue Mar 8 00:46:15 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 8 Mar 2011 00:46:15 +0100 Subject: [GenABEL-dev] changes in MixABEL/src/ITERlib Message-ID: Hi Maksim, I see you have introduced some changes in MixABEL/src/ITERlib in r671. Just to be sure -- is that really intended to improve functionality of MixABEL, or we should roll back? bw, Yurii From yurii.aulchenko at gmail.com Tue Mar 8 00:47:18 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 8 Mar 2011 00:47:18 +0100 Subject: [GenABEL-dev] descriptives.traits by.var bugs. In-Reply-To: <49938.93.48.255.18.1299540121.squirrel@webmail-rupar.regione.fvg.it> References: <53018.212.189.217.13.1299504973.squirrel@webmail-rupar.regione.fvg.it> <49354.93.48.255.18.1299529264.squirrel@webmail-rupar.regione.fvg.it> <49938.93.48.255.18.1299540121.squirrel@webmail-rupar.regione.fvg.it> Message-ID: I would think this is something to solve along the lines of res <- try( operation ); if ( class(res) == "try-error") { ... exception ... } else { ... produce regular results ... } Yurii On Tue, Mar 8, 2011 at 12:22 AM, wrote: > Sorry, I took a better look and the problem and it is sligthly different, > in fact it's not the script but the dataset. > Basically all of the controls are missing the variable age and the t.test > crushes and gives out that error. > I think a check should be added to avoid this from happening, meanwhile > I've posted a reply on the forum on how to get results anyways. > > Best > > Nicola From yurii.aulchenko at gmail.com Tue Mar 8 12:41:22 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 8 Mar 2011 12:41:22 +0100 Subject: [GenABEL-dev] [Genabel-commits] r691 - pkg/ABEL-general/scripts In-Reply-To: <20110308111840.6CAAA9CE3E@r-forge.r-project.org> References: <20110308111840.6CAAA9CE3E@r-forge.r-project.org> Message-ID: Complete diff @ http://lists.r-forge.r-project.org/pipermail/genabel-commits/2011-March/000074.html > +rm -rf VariABEL* > +rm -rf DatABEL DatABEL.* > +rm -rf filevector* The "DatABEL.*" presumably takes care of "datABEL.Rcheck" I think just rm -rf DatABEL* should do, if not then also rm -rf VariABEL* should be rm -rf VariABEL* VariABEL.* > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/filevector > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/DatABEL > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/VariABEL ok, this script is maksim-specific (and others are yurii-specific) :) Yurii From yurii.aulchenko at gmail.com Tue Mar 8 12:47:29 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 8 Mar 2011 12:47:29 +0100 Subject: [GenABEL-dev] [Genabel-commits] r691 - pkg/ABEL-general/scripts In-Reply-To: References: <20110308111840.6CAAA9CE3E@r-forge.r-project.org> Message-ID: I should have tald what this code is about. The scripts at pkg/ABEL-general/scripts/makedistrib_MyVerySpecialABEL.sh are building distrib-version of a specific ABEL. This is usually done before submitting a package to CRAN. Basically, these scripts take care of the complicated dir-structure we may have in some *ABELs because (we eg use symlinks to other *ABELs some times). So, they chack out all packages necessary to maintain symlinks right, copy everything to 'src', and then it all compiles well also when CRAN team does this compilation outside our devel-setting. Yurii From l.karssen at erasmusmc.nl Tue Mar 8 12:55:37 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Tue, 08 Mar 2011 12:55:37 +0100 Subject: [GenABEL-dev] [Genabel-commits] r691 - pkg/ABEL-general/scripts In-Reply-To: References: <20110308111840.6CAAA9CE3E@r-forge.r-project.org> Message-ID: <1299585337.7055.22.camel@rubidium> On Tue, 2011-03-08 at 12:41 +0100, Yurii Aulchenko wrote: > Complete diff @ > > http://lists.r-forge.r-project.org/pipermail/genabel-commits/2011-March/000074.html > > > > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/filevector > > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/DatABEL > > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/VariABEL > > ok, this script is maksim-specific (and others are yurii-specific) :) You could replace maksim with ${USER}. As long as the r-forge user name is equal to that on the system where the shell script is run it 'll work. Lennart. > > Yurii > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 email: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From yurii.aulchenko at gmail.com Tue Mar 8 13:56:25 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 8 Mar 2011 13:56:25 +0100 Subject: [GenABEL-dev] [Genabel-commits] r691 - pkg/ABEL-general/scripts In-Reply-To: <1299585337.7055.22.camel@rubidium> References: <20110308111840.6CAAA9CE3E@r-forge.r-project.org> <1299585337.7055.22.camel@rubidium> Message-ID: >> > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/filevector >> > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/DatABEL >> > +svn export svn+ssh://maksim at svn.r-forge.r-project.org/svnroot/genabel/pkg/VariABEL >> >> ok, this script is maksim-specific (and others are yurii-specific) :) > > You could replace maksim with ${USER}. As long as the r-forge user name > is equal to that on the system where the shell script is run it 'll > work. > Would not work for me -- I have different login names. What about passing a command line arg? should be relatively easy... Yurii From yurii.aulchenko at gmail.com Wed Mar 9 10:49:44 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 9 Mar 2011 10:49:44 +0100 Subject: [GenABEL-dev] [Genabel-commits] r671 - pkg/MixABEL/src/ITERlib In-Reply-To: <20110307104614.C38E49CE46@r-forge.r-project.org> References: <20110307104614.C38E49CE46@r-forge.r-project.org> Message-ID: Hi Maksim, did you really intend to do these changes to MixABEL, or there is some mistake here? I have checked out the version after these modifications, and it does not compile anymore: blah-blah-blah g++ -I/Library/Frameworks/R.framework/Resources/include -I/usr/local/include -fopenmp -fPIC -g -O2 -c iterator_functions.cpp -o iterator_functions.o iterator_functions.cpp:3:35: error: var_homogeneity_tests.h: No such file or directory blah-blah-blah Please comment! Yurii On Mon, Mar 7, 2011 at 11:46 AM, wrote: > Author: maksim > Date: 2011-03-07 11:46:14 +0100 (Mon, 07 Mar 2011) > New Revision: 671 > > Modified: > ? pkg/MixABEL/src/ITERlib/iterator.cpp > ? pkg/MixABEL/src/ITERlib/iterator_functions.cpp > ? pkg/MixABEL/src/ITERlib/iterator_functions.h > Log: > full log at http://lists.r-forge.r-project.org/pipermail/genabel-commits/2011-March/000054.html From m.struchalin at erasmusmc.nl Wed Mar 9 12:40:04 2011 From: m.struchalin at erasmusmc.nl (Maksim Struchalin) Date: Wed, 09 Mar 2011 12:40:04 +0100 Subject: [GenABEL-dev] [Genabel-commits] r671 - pkg/MixABEL/src/ITERlib In-Reply-To: References: <20110307104614.C38E49CE46@r-forge.r-project.org> Message-ID: <4D776714.1070701@erasmusmc.nl> Hi Yurii, All my changes in MixABEL are reverted back. Sorry for disturbance. best, Maksim On 09-03-2011 10:49, Yurii Aulchenko wrote: > Hi Maksim, > > did you really intend to do these changes to MixABEL, or there is some > mistake here? > > I have checked out the version after these modifications, and it does > not compile anymore: > > blah-blah-blah > g++ -I/Library/Frameworks/R.framework/Resources/include > -I/usr/local/include -fopenmp -fPIC -g -O2 -c > iterator_functions.cpp -o iterator_functions.o > iterator_functions.cpp:3:35: error: var_homogeneity_tests.h: No such > file or directory > blah-blah-blah > > Please comment! > > Yurii > > On Mon, Mar 7, 2011 at 11:46 AM, wrote: >> Author: maksim >> Date: 2011-03-07 11:46:14 +0100 (Mon, 07 Mar 2011) >> New Revision: 671 >> >> Modified: >> pkg/MixABEL/src/ITERlib/iterator.cpp >> pkg/MixABEL/src/ITERlib/iterator_functions.cpp >> pkg/MixABEL/src/ITERlib/iterator_functions.h >> Log: >> > full log at > > http://lists.r-forge.r-project.org/pipermail/genabel-commits/2011-March/000054.html From yurii.aulchenko at gmail.com Wed Mar 9 12:51:19 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 9 Mar 2011 12:51:19 +0100 Subject: [GenABEL-dev] [Genabel-commits] r671 - pkg/MixABEL/src/ITERlib In-Reply-To: <4D776714.1070701@erasmusmc.nl> References: <20110307104614.C38E49CE46@r-forge.r-project.org> <4D776714.1070701@erasmusmc.nl> Message-ID: Thanks, Maksim, This now exactly the same as my local version -- so I think we are having right code for MixA now :) Does compile. Yurii On Wed, Mar 9, 2011 at 12:40 PM, Maksim Struchalin wrote: > Hi Yurii, > > All my changes in MixABEL are reverted back. > Sorry for disturbance. > > best, > Maksim From kianvl at mail.ru Wed Mar 9 17:33:00 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Wed, 09 Mar 2011 19:33:00 +0300 Subject: [GenABEL-dev] =?utf-8?q?web-server_todo=27s?= In-Reply-To: References:

Message-ID: I checked http://test.genabel.org/drupal6/?q=packages 1. Links /drupal6/?q=... was changed to ?q=... 2. For ParallABEL and MixABEL there was no links to manual. Checked. 3. Wrong links: 3.1. source code of MixABEL and ProbABEL; 3.2. manuals in pdf for MixABEL and ProbABEL; 3.3. manuals in html for all packages. Do we have the final version of html-manuals? Anatoly Wed, 9 Mar 2011 15:40:28 +0100 ?????? ?? Yurii Aulchenko : > sure -- you mentioned that you will only be able to work on Wed (today) > > by the end of today / start of tomorrow we actually need everything done > > Please use "reply to all" so people do know what is going on > > On Wed, Mar 9, 2011 at 2:53 PM, Anatoliy Kirichenko wrote: > >> > 1) re-structure 'Packages' -- separate sub-page for every package > >> > >> done! Anatoly/Lennart -- can you please have a look if I filled in > >> links all right (I was trying to make relative references). Let me > >> know if I should modify my behavior! > >> > >> > 2) generate HTML manuals for ParallABEL and PredictABEL and put these > >> > to http://test.genabel.org/old/ > >> > >> done! Now, Anatoly -- this is your move. We need all this htmls > >> imported and listed under the 'manuals' ?node > > http://test.genabel.org/old/ - The requested URL /old/ was not found on this > server. > > > >> > >> That's it for today. I have a feeling I am talking a bit to myself -- > >> which I do :P > > I was out. > > > > > > Anatoly > > > > From l.karssen at erasmusmc.nl Wed Mar 9 18:47:06 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Wed, 09 Mar 2011 18:47:06 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References:

Message-ID: <1299692826.15383.2.camel@rubidium> On Wed, 2011-03-09 at 19:33 +0300, Anatoliy Kirichenko wrote: > I checked > http://test.genabel.org/drupal6/?q=packages > > 1. Links /drupal6/?q=... was changed to ?q=... > > 2. For ParallABEL and MixABEL there was no links to manual. Checked. > > 3. Wrong links: > 3.1. source code of MixABEL and ProbABEL; > 3.2. manuals in pdf for MixABEL and ProbABEL; > 3.3. manuals in html for all packages. > > Do we have the final version of html-manuals? Yes, their are on the local file system of the server, in /var/www/old/ (exactly the default used by import_html). $ ls old htmlDatA htmlMetA htmlParallA htmlTut ProbABEL_manual.html tmp htmlGenA htmlMixA htmlPredictA pdfs software Lennart. > > Anatoly > > > Wed, 9 Mar 2011 15:40:28 +0100 ?????? ?? Yurii Aulchenko : > > > sure -- you mentioned that you will only be able to work on Wed (today) > > > > by the end of today / start of tomorrow we actually need everything done > > > > Please use "reply to all" so people do know what is going on > > > > On Wed, Mar 9, 2011 at 2:53 PM, Anatoliy Kirichenko wrote: > > >> > 1) re-structure 'Packages' -- separate sub-page for every package > > >> > > >> done! Anatoly/Lennart -- can you please have a look if I filled in > > >> links all right (I was trying to make relative references). Let me > > >> know if I should modify my behavior! > > >> > > >> > 2) generate HTML manuals for ParallABEL and PredictABEL and put these > > >> > to http://test.genabel.org/old/ > > >> > > >> done! Now, Anatoly -- this is your move. We need all this htmls > > >> imported and listed under the 'manuals' node > > > http://test.genabel.org/old/ - The requested URL /old/ was not found on this > > server. > > > > > >> > > >> That's it for today. I have a feeling I am talking a bit to myself -- > > >> which I do :P > > > I was out. > > > > > > > > > Anatoly > > > > > > > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 email: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From l.karssen at erasmusmc.nl Wed Mar 9 19:16:47 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Wed, 09 Mar 2011 19:16:47 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References:

Message-ID: <1299694607.15383.4.camel@rubidium> On Wed, 2011-03-09 at 19:33 +0300, Anatoliy Kirichenko wrote: > I checked > http://test.genabel.org/drupal6/?q=packages > > 1. Links /drupal6/?q=... was changed to ?q=... > > 2. For ParallABEL and MixABEL there was no links to manual. Checked. There are not any manuals for those packages. So for now they need to be greyed out or something. > > 3. Wrong links: > 3.1. source code of MixABEL and ProbABEL; We are trying to fix this. For some reason the URL test.genabel.org/old doesn't work any more. We are looking into this at the moment. Lennart and Yurii. > 3.2. manuals in pdf for MixABEL and ProbABEL; > 3.3. manuals in html for all packages. > > Do we have the final version of html-manuals? > > Anatoly > > > Wed, 9 Mar 2011 15:40:28 +0100 ?????? ?? Yurii Aulchenko : > > > sure -- you mentioned that you will only be able to work on Wed (today) > > > > by the end of today / start of tomorrow we actually need everything done > > > > Please use "reply to all" so people do know what is going on > > > > On Wed, Mar 9, 2011 at 2:53 PM, Anatoliy Kirichenko wrote: > > >> > 1) re-structure 'Packages' -- separate sub-page for every package > > >> > > >> done! Anatoly/Lennart -- can you please have a look if I filled in > > >> links all right (I was trying to make relative references). Let me > > >> know if I should modify my behavior! > > >> > > >> > 2) generate HTML manuals for ParallABEL and PredictABEL and put these > > >> > to http://test.genabel.org/old/ > > >> > > >> done! Now, Anatoly -- this is your move. We need all this htmls > > >> imported and listed under the 'manuals' node > > > http://test.genabel.org/old/ - The requested URL /old/ was not found on this > > server. > > > > > >> > > >> That's it for today. I have a feeling I am talking a bit to myself -- > > >> which I do :P > > > I was out. > > > > > > > > > Anatoly > > > > > > > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 email: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From yurii.aulchenko at gmail.com Wed Mar 9 20:13:07 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 9 Mar 2011 20:13:07 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: <1299694607.15383.4.camel@rubidium> References:

<1299694607.15383.4.camel@rubidium> Message-ID: >> 2. For ParallABEL and MixABEL there was no links to manual. Checked. > There are not any manuals for those packages. So for now they need to be > greyed out or something. > sorry guys: the MANUALS are there (/var/www/old/...). But not TUTORIALS (tut are only available for GenA, MetA -> general ABEL tutotrial; DatA -- specific vignette). Yurii From yurii.aulchenko at gmail.com Wed Mar 9 22:12:18 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 9 Mar 2011 22:12:18 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References:

Message-ID: Anatoly et al., I have been patching web-pages according to your suggestions, and also updated the content a bit. All changes submitted to drupal pages as well. Hopefully all links are in right format. still todo: 1) Anatoly: manuals section now contains only DatABEL manual, there are 6 other manuals to be imported 2) Lennart: we have some packages and manuals-PDF at /old/ -- this does not work yet 3) Lennart: migrate to www.genabel.org > 1. Links /drupal6/?q=... was changed to ?q=... > addressed this across all pages > 2. For ParallABEL and MixABEL there was no links to manual. Checked. addressed > > 3. Wrong links: > 3.1. source code of MixABEL and ProbABEL; > 3.2. manuals in pdf for MixABEL and ProbABEL; addressed, but depends upon 2) > 3.3. manuals in html for all packages. they are there, as described in e-mail of Lennart > > Do we have the final version of html-manuals? > yes (e-mail of Lennart) Thanks a lot! It seems we can make it by the deadline! Yurii On Wed, Mar 9, 2011 at 10:01 PM, wrote: > Author: yurii > Date: 2011-03-09 22:01:14 +0100 (Wed, 09 Mar 2011) > New Revision: 696 > > Modified: > pkg/webpages/acknowledgements.html > pkg/webpages/contribute.html > pkg/webpages/home.html > pkg/webpages/packages.html > pkg/webpages/people.html > pkg/webpages/pkgMixABEL.html > pkg/webpages/pkgProbABEL.html > pkg/webpages/showcase.html > pkg/webpages/support.html > pkg/webpages/tutorials.html > Log: > webpages: fixing links, patching the content > From yurii.aulchenko at gmail.com Wed Mar 9 22:17:38 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 9 Mar 2011 22:17:38 +0100 Subject: [GenABEL-dev] dealing with suggestions on the devel-list Message-ID: Dear All, I expect with more people looking at our code we will get more and more suggestions on how to fix a bug (e.g. Nicola's recent post on treating 'by.var' problem); we also may expect people posting patches (this has happened before, when genabel-devel list was not there yet). It is probably good time to think how we deal with these situations. What I could think of looks very straightforward to me, but let us know if you think different * Situation 1. We get a suggestion. a) discuss the suggestion on the list b) if suggestion passes the discussion, a committer submits outline of bug fix to the bug tracker c) a committer "assigns" bug to himself (bug tracker) d) the commiter implements the changes, tests, commit, ... (usual route) ("a committer" refers to anyone who have volunteered, though we understand that for the near future this will likely be one of the project coordinators) * Situation 2. We get a patch. a) discuss the patch on the list b) if patch passes the discussion, a committer "assigns" bug to himself (bug tracker) c) the committer patch the code, run checks, commit, ... (usual route) Let me know what you think of the proposed procedure. best wishes, Yurii From l.karssen at erasmusmc.nl Thu Mar 10 00:58:21 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Thu, 10 Mar 2011 00:58:21 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References:

Message-ID: <1299715101.2466.73.camel@sidonia> Dear all, On Wed, 2011-03-09 at 22:12 +0100, Yurii Aulchenko wrote: > still todo: > 2) Lennart: we have some packages and manuals-PDF at /old/ -- this > does not work yet See next point. 'old' works again. > > 3) Lennart: migrate to www.genabel.org Done. www.genabel.org/ points to drupal. forum.genabel.org points to the forum. Neither of these is reachable anymore via their old (second) URL (test.genabel.org/drupal6/ and test.genabel.org/phpbb/, respectively. This way we remove possible confusion over the correct URL to each of these sites (also for search engines). The 'old' directory is now only available through test.genabel.org/old/. We should discuss a better name tomorrow it we like to keep this directory around. I also enabled the clean URLs option in drupal (which we couldn't get to work before). The crazy thing is that now drupal URLs like www.genabel.org/anything_here/?q=people work... As long as the page after ?q= actually exists you get pointed to the right page. A side effect I'd say, but as a result it is not strictly necessary to change links from /drupal6/?q= to ?q= (see below)... Actually, with the clean URLs module enabled, the links should now be changed from /drupal6/?q=some_page to /some_page. > > > 1. Links /drupal6/?q=... was changed to ?q=... Actually it should be /?q=... (or only ... as mentioned in the previous point). However, as described, the way drupal is set up now is very forgiving for these things. Lennart. > > > addressed this across all pages > > > 2. For ParallABEL and MixABEL there was no links to manual. Checked. > > addressed > > > > > Thanks a lot! It seems we can make it by the deadline! > > Yurii > > > > On Wed, Mar 9, 2011 at 10:01 PM, wrote: > > Author: yurii > > Date: 2011-03-09 22:01:14 +0100 (Wed, 09 Mar 2011) > > New Revision: 696 > > > > Modified: > > pkg/webpages/acknowledgements.html > > pkg/webpages/contribute.html > > pkg/webpages/home.html > > pkg/webpages/packages.html > > pkg/webpages/people.html > > pkg/webpages/pkgMixABEL.html > > pkg/webpages/pkgProbABEL.html > > pkg/webpages/showcase.html > > pkg/webpages/support.html > > pkg/webpages/tutorials.html > > Log: > > webpages: fixing links, patching the content > > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee 2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 e-mail: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From yurii.aulchenko at gmail.com Thu Mar 10 01:14:31 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 10 Mar 2011 01:14:31 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: <1299715101.2466.73.camel@sidonia> References:

<1299715101.2466.73.camel@sidonia> Message-ID: Lennart, this is terrific! We are really making it towards the deadline! One thing left for Anatoly -- manuals from Apache /old -> import and file in into /?q=manuals, check links... (now, I really like that fact that Anatoly is +5 hours from CET :) )... And then we can ask people to have a look! After that still todo on March 14 or so: old MGA pages = redirect to new pages (Anatoly, can you please file this into your queue?) thanks a lot, and good night! Yurii On Thu, Mar 10, 2011 at 12:58 AM, L.C. Karssen wrote: > Dear all, > > On Wed, 2011-03-09 at 22:12 +0100, Yurii Aulchenko wrote: > >> still todo: > >> 2) Lennart: we have some packages and manuals-PDF at /old/ -- this >> does not work yet > See next point. 'old' works again. > >> >> 3) Lennart: migrate to www.genabel.org > > Done. www.genabel.org/ points to drupal. forum.genabel.org points to the > forum. Neither of these is reachable anymore via their old (second) URL > (test.genabel.org/drupal6/ and test.genabel.org/phpbb/, respectively. > This way we remove possible confusion over the correct URL to each of > these sites (also for search engines). > > The 'old' directory is now only available through test.genabel.org/old/. > We should discuss a better name tomorrow it we like to keep this > directory around. > > I also enabled the clean URLs option in drupal (which we couldn't get to > work before). > The crazy thing is that now drupal URLs like > www.genabel.org/anything_here/?q=people work... As long as the page > after ?q= actually exists you get pointed to the right page. A side > effect I'd say, but as a result it is not strictly necessary to change > links from /drupal6/?q= to ?q= (see below)... Actually, with the clean > URLs module enabled, the links should now be changed > from /drupal6/?q=some_page to /some_page. > > >> >> > 1. Links /drupal6/?q=... was changed to ?q=... > Actually it should be /?q=... (or only ... as mentioned in the previous > point). However, as described, the way drupal is set up now is very > forgiving for these things. > > > Lennart. > >> > >> addressed this across all pages >> >> > 2. For ParallABEL and MixABEL there was no links to manual. Checked. >> >> addressed >> >> > >> >> Thanks a lot! It seems we can make it by the deadline! >> >> Yurii >> >> >> >> On Wed, Mar 9, 2011 at 10:01 PM, ? wrote: >> > Author: yurii >> > Date: 2011-03-09 22:01:14 +0100 (Wed, 09 Mar 2011) >> > New Revision: 696 >> > >> > Modified: >> > ? pkg/webpages/acknowledgements.html >> > ? pkg/webpages/contribute.html >> > ? pkg/webpages/home.html >> > ? pkg/webpages/packages.html >> > ? pkg/webpages/people.html >> > ? pkg/webpages/pkgMixABEL.html >> > ? pkg/webpages/pkgProbABEL.html >> > ? pkg/webpages/showcase.html >> > ? pkg/webpages/support.html >> > ? pkg/webpages/tutorials.html >> > Log: >> > webpages: fixing links, patching the content >> > >> _______________________________________________ >> genabel-devel mailing list >> genabel-devel at lists.r-forge.r-project.org >> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > > -- > ----------------------------------------------- > L.C. Karssen > Erasmus MC > Department of Epidemiology > Room Ee 2224 > > Postbus 2040 > 3000 CA Rotterdam > The Netherlands > > phone: +31-10-7044217 > fax: +31-10-7044657 > e-mail: l.karssen at erasmusmc.nl > GPG key ID: 0E1D39E3 > ----------------------------------------------- > From kianvl at mail.ru Thu Mar 10 05:24:45 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Thu, 10 Mar 2011 07:24:45 +0300 Subject: [GenABEL-dev] =?utf-8?q?web-server_todo=27s?= In-Reply-To: References:

Message-ID: > 3) Lennart: migrate to www.genabel.org Great! Now I'll make all links clear relative! Anatoly From kianvl at mail.ru Thu Mar 10 07:19:53 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Thu, 10 Mar 2011 09:19:53 +0300 Subject: [GenABEL-dev] =?utf-8?q?web-server_todo=27s?= In-Reply-To: References:

Message-ID: All internal links was changed into clear relative links: /drupal6/?q=/blah-blah-blah -> /blah-blah-blah ?q=/blah-blah-blah -> /blah-blah-blah I offer to rename /htmlDatA into /DatABEL Anatoly From kianvl at mail.ru Thu Mar 10 10:00:27 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Thu, 10 Mar 2011 12:00:27 +0300 Subject: [GenABEL-dev] =?utf-8?q?web-server_todo=27s?= In-Reply-To: References: <1299715101.2466.73.camel@sidonia>

Message-ID: I am copying manuals into web-page. There is no some resources: ../DESCRIPTION ../../../doc/html/logo.jpg Anatoly From yurii.aulchenko at gmail.com Thu Mar 10 10:23:14 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 10 Mar 2011 10:23:14 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References: <1299715101.2466.73.camel@sidonia>

Message-ID: Yes, this is a bit of problem indeed, but not a big one. I suggest we just ignore it for the moment (grey out links? Or really ignore? Whatever is easier). Basically these htmls are taken from the html-doc generated by R, and it includes upper levels as well, etc. We can try to see if we still could have "closed form" for separate packages. Yurii On 3/10/11, Anatoliy Kirichenko wrote: > I am copying manuals into web-page. > There is no some resources: > ../DESCRIPTION > ../../../doc/html/logo.jpg > > Anatoly > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > -- Sent from my mobile device From kianvl at mail.ru Thu Mar 10 10:26:42 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Thu, 10 Mar 2011 12:26:42 +0300 Subject: [GenABEL-dev] =?utf-8?q?web-server_todo=27s?= In-Reply-To: References: <1299715101.2466.73.camel@sidonia> Message-ID: > Yes, this is a bit of problem indeed, but not a big one. I suggest we > just ignore it for the moment (grey out links? Or really ignore? > Whatever is easier). As I understand, the links are not informative. Let's remove them permanently? > > Basically these htmls are taken from the html-doc generated by R, and > it includes upper levels as well, etc. We can try to see if we still > could have "closed form" for separate packages. > > Yurii > > On 3/10/11, Anatoliy Kirichenko wrote: > > I am copying manuals into web-page. > > There is no some resources: > > ../DESCRIPTION > > ../../../doc/html/logo.jpg > > > > Anatoly > > > > _______________________________________________ > > genabel-devel mailing list > > genabel-devel at lists.r-forge.r-project.org > > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > > > > -- > Sent from my mobile device From kianvl at mail.ru Thu Mar 10 16:53:24 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Thu, 10 Mar 2011 18:53:24 +0300 Subject: [GenABEL-dev] Manuals In-Reply-To: References:

Message-ID: All manuals in html format are available on http://www.genabel.org/manuals Anatoly From yurii.aulchenko at gmail.com Thu Mar 10 17:08:18 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 10 Mar 2011 17:08:18 +0100 Subject: [GenABEL-dev] Manuals In-Reply-To: References:

Message-ID: Many thanks, Anatoly, for the hard job around the clock! have a very good night :) We now can start asking people to have a look! On Thu, Mar 10, 2011 at 4:53 PM, Anatoliy Kirichenko wrote: > All manuals in html format are available on > http://www.genabel.org/manuals > > Anatoly > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > From pirastu at burlo.trieste.it Fri Mar 11 09:50:56 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Fri, 11 Mar 2011 09:50:56 +0100 (CET) Subject: [GenABEL-dev] dealing with suggestions on the devel-list In-Reply-To: References: Message-ID: <49485.212.189.217.13.1299833456.squirrel@webmail-rupar.regione.fvg.it> Dear all, I think that the procedure seems very straight foward and resonable, for patches we might want to outline a readme wich should be released with the patch with some things like what varsion of the GenABEL-dev or CRAN it was created with and what it does. Nicola > * Situation 1. We get a suggestion. > a) discuss the suggestion on the list > b) if suggestion passes the discussion, a committer submits outline of > bug fix to the bug tracker > c) a committer "assigns" bug to himself (bug tracker) > d) the commiter implements the changes, tests, commit, ... (usual route) > ("a committer" refers to anyone who have volunteered, though we > understand that for the near future this will likely be one of the > project coordinators) >* Situation 2. We get a patch. > a) discuss the patch on the list > b) if patch passes the discussion, a committer "assigns" bug to > himself (bug tracker) > c) the committer patch the code, run checks, commit, ... (usual route) > Let me know what you think of the proposed procedure. > best wishes, > Yurii From yurii.aulchenko at gmail.com Fri Mar 11 10:31:53 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Fri, 11 Mar 2011 10:31:53 +0100 Subject: [GenABEL-dev] content of www.genabel.org Message-ID: Dear All, If you are subscribed to the genabel-development mailing list, you may have noticed that during the last few days Anatoly and Lennart did a lot of work on getting our web-site into the right technical state. Now is the time to think about the content. I have filled in some of the content according to my view of our project; however, our web-site should reflect our consensus view as "GenABEL project" community. Therefore, please have a look, and reply to this mailing list. Any kind of suggestions, comments, and criticism are more than welcome. Also please let us know if I forgot to mention something (or someone!). This web-site will be the public face of the project, serving as a hub connecting users and developers. In that it will play a big role in the dissemination of the results of your efforts. We aim that within the "GenABEL project", every contributor gets proper acknowledgment and recognition. In relation to the web-site, it is now designed in such a way that every package has a separate page, and we would like these to be maintained (and extended!) by the individual contributors. Also, please have a look at the "People" page. The web-site will be 'officially' announced on March 15 in a paper by Cecile Janssens & Suman Kundu (PredictABEL). It would be good to have a consensus on the content and general setup before that. On behalf of the web-site team, Yurii From william.astle at imperial.ac.uk Fri Mar 11 11:27:32 2011 From: william.astle at imperial.ac.uk (William Astle) Date: Fri, 11 Mar 2011 10:27:32 +0000 Subject: [GenABEL-dev] dealing with suggestions on the devel-list In-Reply-To: References: Message-ID: <4D79F914.9090506@ic.ac.uk> On 09/03/11 21:17, Yurii Aulchenko wrote: > Let me know what you think of the proposed procedure. I think these are quite sensible procedures. bw Will. From fabregat at aices.rwth-aachen.de Fri Mar 11 13:52:36 2011 From: fabregat at aices.rwth-aachen.de (Diego Fabregat Traver) Date: Fri, 11 Mar 2011 13:52:36 +0100 Subject: [GenABEL-dev] dealing with suggestions on the devel-list In-Reply-To: References: Message-ID: Dear all, this is my first email to the list so "Hello everyone!". I've never been too involved in the maintenance stage of a software life cycle but the steps suggested by Yurii seem reasonable to me. Which has been the approach so far? Did you already have a bug-tracker? Was the approach successful? If it used to work, I guess there is no need to change it. I think the possible situations are many. There can be bug reports, patches, feature requests, ... something else? And the way to deal with them probably depends on the number of users/reports you get. If the estimated number of reports is low probably it is possible to handle them in the mailing list, but when the number increases maybe dealing with them through the mailing list may become unmanageable. Again, I'm not too experienced on software maintenance and I don't know anything about the way the project is organized, so take my words just as mere thoughts :) Best, Diego El 09/03/11, Yurii Aulchenko escribi?: > Dear All, > > I expect with more people looking at our code we will get more and > more suggestions on how to fix a bug (e.g. Nicola's recent post on > treating 'by.var' problem); we also may expect people posting patches > (this has happened before, when genabel-devel list was not there yet). > It is probably good time to think how we deal with these situations. > > What I could think of looks very straightforward to me, but let us > know if you think different > > * Situation 1. We get a suggestion. > > a) discuss the suggestion on the list > b) if suggestion passes the discussion, a committer submits outline of > bug fix to the bug tracker > c) a committer "assigns" bug to himself (bug tracker) > d) the commiter implements the changes, tests, commit, ... (usual route) > > ("a committer" refers to anyone who have volunteered, though we > understand that for the near future this will likely be one of the > project coordinators) > > * Situation 2. We get a patch. > > a) discuss the patch on the list > b) if patch passes the discussion, a committer "assigns" bug to > himself (bug tracker) > c) the committer patch the code, run checks, commit, ... (usual route) > > Let me know what you think of the proposed procedure. > > best wishes, > Yurii > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel From yurii.aulchenko at gmail.com Mon Mar 14 00:09:05 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 14 Mar 2011 00:09:05 +0100 Subject: [GenABEL-dev] Fwd: People page on the GenABEL.org In-Reply-To: References: Message-ID: FYI ---------- Forwarded message ---------- From: Yurii Aulchenko Date: Mon, Mar 14, 2011 at 12:06 AM Subject: People page on the GenABEL.org To: Yurii Aulchenko Cc: Lennart Karssen, Anatoly Kirichenko Bcc: Dear GenABEL project contributor, You might have received (and already addressed) this information before, in which case please disregard it, and sorry for any double posting. We want to acknowledge your contribution to the GenABEL project by mentioning your name at the project's "People" web-page (http://www.genabel.org/people). The GenABEL project web-site (http://www.genabel.org/home), intended to serve as the project's public face and information hub, will be announced through several channels this Tuesday, March 15, 2011. If you do not want your name to appear on the GenABEL project web-site, please notify us in a reply e-mail as soon as possible. If you feel that your contribution is not adequately described, please suggest changes. You may consider to ask us to add your title and/or affiliation, so that your organization also receives some credit. Please notify us if you think we forget to mention something (or someone). We will also greatly appreciate your comments on the content of GenABEL.org in general. You may want to see genabel-development mail http://lists.r-forge.r-project.org/pipermail/genabel-devel/2011-March/000163.html for a general overview of the envisioned role of GenABEL.org within the GenABEL project. On behalf of the GenABEL web-site development team (cc'ed), Yurii Aulchenko From yurii.aulchenko at gmail.com Mon Mar 14 20:41:24 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 14 Mar 2011 20:41:24 +0100 Subject: [GenABEL-dev] dealing with suggestions on the devel-list In-Reply-To: References:

Message-ID: Dear Diego, many thanks for your suggestions! Here are some comments: > ... steps suggested by Yurii seem reasonable to me. Which has been the approach so far? Did you already have a bug-tracker? Was the approach successful? If it used to work, I guess there is no need to change it. > We do have a bug tracker at https://r-forge.r-project.org/tracker/?group_id=505 but only committers can file bugs in. Up until now, mostly Lennart and me filled these in. Bug reports came from users reporting directly to me (pre-forum & pre genabel-devel list era), or through the genabel-devel mailing list, or through the forum (forum.genabel.org). Reporting through forum actually worked quite fantastically, see example http://forum.genabel.org/viewtopic.php?f=6&t=4#p17 So, this procedure of filing the bugs and feature requests (now mostly through forum) is in place, and it works for now. Now we start getting suggestions on 'how to fix a bug' and in writing my e-mail, I was aiming to discuss how we deal with this anticipated situation, and set our guidelines through the discussion. > ... the way to deal with them probably depends on the number of users/reports you get. If the estimated number of reports is low probably it is possible to handle them in the mailing list, but when the number increases maybe dealing with them through the mailing list may become unmanageable. > This is very good point, and it is also highlights a very important question that number of mailings in the genabel-devel list may get too high, too soon. Bug reports, feature suggestions, suggestions on how to fix a bug, patches; and also we intend to have some methodology discussion! I am going to rise the question how we deal with that in my next e-mail(s). Thanks again for sharing your ideas, Yurii From yurii.aulchenko at gmail.com Mon Mar 14 20:47:52 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 14 Mar 2011 20:47:52 +0100 Subject: [GenABEL-dev] dealing with suggestions on the devel-list In-Reply-To: References: Message-ID: I will try to summarize the discussion of the question "how we deal with suggested bug fixes/patches on the genabel-devel list", or at least describe how this discussion has reflected in my mind. 1) In general, the guidelines suggested look reasonable 2) Nicola suggested that for patches we need a set of guidelines/HOWTO on "reporting a patch". This is a separate topic to follow-up, probably when we get a first patch, we can ask the contributor to draft a HOWTO, and then discuss that and publish; a framework similar to one we used to develop http://genabel.r-forge.r-project.org/tutHowToInstallDevelVersion.html 3) Diego drew our attention to that fact that the framework may be different depending on the number of suggestions we may get. This raises a very important question, which I would like to discuss with you as a separate topic. Please reply if I misinterpreted your suggestions, or if you would like to add something. best wishes, Yurii On Wed, Mar 9, 2011 at 10:17 PM, Yurii Aulchenko wrote: > Dear All, > > I expect with more people looking at our code we will get more and > more suggestions on how to fix a bug (e.g. Nicola's recent post on > treating 'by.var' problem); we also may expect people posting patches > (this has happened before, when genabel-devel list was not there yet). > It is probably good time to think how we deal with these situations. > > What I could think of looks very straightforward to me, but let us > know if you think different > > * Situation 1. We get a suggestion. > > a) discuss the suggestion on the list > b) if suggestion passes the discussion, a committer submits outline of > bug fix to the bug tracker > c) a committer "assigns" bug to himself (bug tracker) > d) the commiter implements the changes, tests, commit, ... (usual route) > > ("a committer" refers to anyone who have volunteered, though we > understand that for the near future this will likely be one of the > project coordinators) > > * Situation 2. We get a patch. > > a) discuss the patch on the list > b) if patch passes the discussion, a committer "assigns" bug to > himself (bug tracker) > c) the committer patch the code, run checks, commit, ... (usual route) > > Let me know what you think of the proposed procedure. > > best wishes, > Yurii > From yurii.aulchenko at gmail.com Mon Mar 14 21:26:42 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Mon, 14 Mar 2011 21:26:42 +0100 Subject: [GenABEL-dev] why genabel-devel list only? (genabel-devel = SPAM?) Message-ID: Dear All, I was looking at the number of e-mails at this (genabel-devel) list, and I was quite happy to see that the number of people posting and number of messages is growing (#mails Jan = 18 / #people = 3; Feb = 61 / 6; mid March = 60 / 9), which I take as reflection of the fact that the project gets bigger, and more active. A couple of really important issues were solved in the process of discussion on the genabel-devel list. So far so good, and congratulations to all of us, but 120 mails per month (projection for March) is 4 mails per day! Now, we are all people busy with many projects, and I can tell for myself that I could not stand most really active mailing-lists: while there are a few posts of high interest, there are also too many posts, which I am not interested to read! So, after a while, I would unsubscribe from such "active" mailing list. I am very afraid this is something, which may happen to genabel-devel list as well. Things like discussion over 'faster polygenic' is of interest to many of us, but not all. The 'web-server' discussion is most interesting to the web-team, but most of others may be willing to have an occasional look and give a suggestion, but are not up to receive, read, and react to 5 posts on this topic per day. This is something we can change for better. My view is that we do need to have our discussions in open, so anyone can have a look, join in, contribute and help, but we all also need the freedom to choose the discussions to follow-up closely, and the discussions we might be willing to have an occasional look at. My suggestion is to stream most of the bigger discussions of technical aspects (such as 'web-server') and methodology (such as 'faster polygenic') to a forum -- e.g. we can arrange GenABEL-devel forum, along with GenABEL(-user) forum already in place at forum.genabel.org. Then, anyone will have a freedom to subscribe to particular topic(s), or just follow the whole forum. On the genable-devel list we can announce new topics, and also keep it as a point of contact with users/contributors (bug reports, suggestions, ... -- see http://www.genabel.org/contribute). If we do this I believe we will have targeted information delivery to all contributors, and still will keep the process open. Please let me know what you think, especially of you feel this is not good way to go. with best wishes, Yurii From fabregat at aices.rwth-aachen.de Tue Mar 15 23:50:45 2011 From: fabregat at aices.rwth-aachen.de (Diego Fabregat Traver) Date: Tue, 15 Mar 2011 23:50:45 +0100 Subject: [GenABEL-dev] why genabel-devel list only? (genabel-devel = SPAM?) In-Reply-To: References: Message-ID: Hi Yurii, in our opinion, the forum is good for all the reasons you mentioned. On the other hand, it is not at all as immediate as receiving an email in your mailbox. It's a matter of balance. For users' questions/answers the forum is clearly superior. And the messages stay. Are searchable. For developers, if organized in small groups, a mailing list is more effective. Best, Diego/Paolo El 14/03/11, Yurii Aulchenko escribi?: > Dear All, > > I was looking at the number of e-mails at this (genabel-devel) list, > and I was quite happy to see that the number of people posting and > number of messages is growing (#mails Jan = 18 / #people = 3; Feb = 61 > / 6; mid March = 60 / 9), which I take as reflection of the fact that > the project gets bigger, and more active. A couple of really important > issues were solved in the process of discussion on the genabel-devel > list. > > So far so good, and congratulations to all of us, but 120 mails per > month (projection for March) is 4 mails per day! Now, we are all > people busy with many projects, and I can tell for myself that I could > not stand most really active mailing-lists: while there are a few > posts of high interest, there are also too many posts, which I am not > interested to read! So, after a while, I would unsubscribe from such > "active" mailing list. > > I am very afraid this is something, which may happen to genabel-devel > list as well. Things like discussion over 'faster polygenic' is of > interest to many of us, but not all. The 'web-server' discussion is > most interesting to the web-team, but most of others may be willing to > have an occasional look and give a suggestion, but are not up to > receive, read, and react to 5 posts on this topic per day. > > This is something we can change for better. My view is that we do need > to have our discussions in open, so anyone can have a look, join in, > contribute and help, but we all also need the freedom to choose the > discussions to follow-up closely, and the discussions we might be > willing to have an occasional look at. > > My suggestion is to stream most of the bigger discussions of technical > aspects (such as 'web-server') and methodology (such as 'faster > polygenic') to a forum -- e.g. we can arrange GenABEL-devel forum, > along with GenABEL(-user) forum already in place at forum.genabel.org. > Then, anyone will have a freedom to subscribe to particular topic(s), > or just follow the whole forum. On the genable-devel list we can > announce new topics, and also keep it as a point of contact with > users/contributors (bug reports, suggestions, ... -- see > http://www.genabel.org/contribute). If we do this I believe we will > have targeted information delivery to all contributors, and still will > keep the process open. > > Please let me know what you think, especially of you feel this is not > good way to go. > > with best wishes, > Yurii > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel From l.karssen at erasmusmc.nl Wed Mar 16 08:58:56 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Wed, 16 Mar 2011 08:58:56 +0100 Subject: [GenABEL-dev] why genabel-devel list only? (genabel-devel = SPAM?) In-Reply-To: References:

Message-ID: <1300262336.24076.10.camel@sidonia> Hi all, I agree with the message of Diego and Paolo below: I prefer to keep the mailing list for developer discussions. A mailing list, for example allows one to reply not only to the latest 'post' but also to previous ones, making the 'conversation' less linear and more flexible. Also, those who feel that the devel list has too much traffic and/or are too distracted by a (more or less) constant stream of e-mails, can change their mailing list subscription to the daily digest. Regards, Lennart. On Tue, 2011-03-15 at 23:50 +0100, Diego Fabregat Traver wrote: > Hi Yurii, > > in our opinion, the forum is good for all the reasons you mentioned. > On the other hand, it is not at all as immediate as receiving an email > in your mailbox. It's a matter of balance. For users' > questions/answers the forum is clearly superior. And the messages > stay. Are searchable. For developers, if organized in small groups, a > mailing list is more effective. > > Best, > Diego/Paolo > > > El 14/03/11, Yurii Aulchenko escribi?: > > > Dear All, > > > > I was looking at the number of e-mails at this (genabel-devel) list, > > and I was quite happy to see that the number of people posting and > > number of messages is growing (#mails Jan = 18 / #people = 3; Feb = 61 > > / 6; mid March = 60 / 9), which I take as reflection of the fact that > > the project gets bigger, and more active. A couple of really important > > issues were solved in the process of discussion on the genabel-devel > > list. > > > > So far so good, and congratulations to all of us, but 120 mails per > > month (projection for March) is 4 mails per day! Now, we are all > > people busy with many projects, and I can tell for myself that I could > > not stand most really active mailing-lists: while there are a few > > posts of high interest, there are also too many posts, which I am not > > interested to read! So, after a while, I would unsubscribe from such > > "active" mailing list. > > > > I am very afraid this is something, which may happen to genabel-devel > > list as well. Things like discussion over 'faster polygenic' is of > > interest to many of us, but not all. The 'web-server' discussion is > > most interesting to the web-team, but most of others may be willing to > > have an occasional look and give a suggestion, but are not up to > > receive, read, and react to 5 posts on this topic per day. > > > > This is something we can change for better. My view is that we do need > > to have our discussions in open, so anyone can have a look, join in, > > contribute and help, but we all also need the freedom to choose the > > discussions to follow-up closely, and the discussions we might be > > willing to have an occasional look at. > > > > My suggestion is to stream most of the bigger discussions of technical > > aspects (such as 'web-server') and methodology (such as 'faster > > polygenic') to a forum -- e.g. we can arrange GenABEL-devel forum, > > along with GenABEL(-user) forum already in place at forum.genabel.org. > > Then, anyone will have a freedom to subscribe to particular topic(s), > > or just follow the whole forum. On the genable-devel list we can > > announce new topics, and also keep it as a point of contact with > > users/contributors (bug reports, suggestions, ... -- see > > http://www.genabel.org/contribute). If we do this I believe we will > > have targeted information delivery to all contributors, and still will > > keep the process open. > > > > Please let me know what you think, especially of you feel this is not > > good way to go. > > > > with best wishes, > > Yurii > > _______________________________________________ > > genabel-devel mailing list > > genabel-devel at lists.r-forge.r-project.org > > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee 2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 e-mail: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From yurii.aulchenko at gmail.com Wed Mar 16 18:24:03 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Wed, 16 Mar 2011 18:24:03 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References: <1299504890.1957.196.camel@rubidium> Message-ID: Dear Anatoly, dear All, apparently our web-site is in good shape -- at least there were no reports of serious problems, and we gave few days for the people to react. Unless someone pinpoints a big trouble (please notify immediately!), I suggest that we >>> * Replace MGA *ABEL pages with something, which re-directs the traffic >>> to www.genabel.org Anatoly, can you please take care of this some time this week, if possible? I presume we want to redirect traffic from http://mga.bionet.nsc.ru/nlru/GenABEL/ http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ http://mga.bionet.nsc.ru/~yurii/ABEL/ and (ideally) all other content in these directories. This is important: e.g. if you search google for "ABEL tutorial" you get http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ABEL-tutorial.pdf as the top hit. Just to remind what we had in our discussion about re-direction: "Good" procedure: >> >> This depends a bit on the level of access that you have on MGA. IIRC you >> have the files in your ~/public_html dir, right? There are a couple of >> ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite >> enabled we can use that. See e.g. >> http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an >> example. >> > Simple procedure (but we will be penalized by search engines): > something more simple, which I used for old-old-old > page of GenABEL, see > > http://mga.bionet.nsc.ru/nlru/GenABEL > > basically, it boils down to this HTML line > > > From kianvl at mail.ru Thu Mar 17 05:57:43 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Thu, 17 Mar 2011 07:57:43 +0300 Subject: [GenABEL-dev] =?utf-8?q?web-server_todo=27s?= In-Reply-To: References: Message-ID: Dear Yurii! > Dear Anatoly, dear All, > > apparently our web-site is in good shape -- at least there were no > reports of serious problems, and we gave few days for the people to > react. Unless someone pinpoints a big trouble (please notify > immediately!), I suggest that we > > >>> * Replace MGA *ABEL pages with something, which re-directs the traffic > >>> to www.genabel.org > > Anatoly, can you please take care of this some time this week, if > possible? I presume we want to redirect traffic from > > http://mga.bionet.nsc.ru/nlru/GenABEL/ > http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ > http://mga.bionet.nsc.ru/~yurii/ABEL/ I changed index.html file on this pages with It works. > > and (ideally) all other content in these directories. This is > important: e.g. if you search google for "ABEL tutorial" you get > > http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ABEL-tutorial.pdf The page is redirected too. But there is no pdf-file http://www.genabel.org/old/pdfs/ABEL-tutorial.pdf Which pages should be redirected else? Do you like this way of redirection? > > as the top hit. > > Just to remind what we had in our discussion about re-direction: > > "Good" procedure: > > >> > >> This depends a bit on the level of access that you have on MGA. IIRC you > >> have the files in your ~/public_html dir, right? There are a couple of > >> ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite > >> enabled we can use that. See e.g. > >> http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an > >> example. > >> > > > > Simple procedure (but we will be penalized by search engines): > > > something more simple, which I used for old-old-old > > page of GenABEL, see > > > > http://mga.bionet.nsc.ru/nlru/GenABEL > > > > basically, it boils down to this HTML line > > > > Anatoly From yurii.aulchenko at gmail.com Thu Mar 17 09:39:36 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 17 Mar 2011 09:39:36 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References:

Message-ID: Hi Tolya! Thanks a lot! If you got some time - can you please try "301" solution? As Lennart noticed in previous discussion, we may be "punished" by search engines if use "simple" solution, which is in place now. That pdf: this is ABEL-tutorial, so redirect should go to www.genabel.org/tutorials/ABEL-tutorial or something like this. Should be easy to spot - we have only two tutorials :) We also have PDFs there, but I would rather say we redirect to HTML version. Other files: contact me directly if not clear where redirect should go to. Yurii On 3/17/11, Anatoliy Kirichenko wrote: > Dear Yurii! > >> Dear Anatoly, dear All, >> >> apparently our web-site is in good shape -- at least there were no >> reports of serious problems, and we gave few days for the people to >> react. Unless someone pinpoints a big trouble (please notify >> immediately!), I suggest that we >> >> >>> * Replace MGA *ABEL pages with something, which re-directs the traffic >> >>> to www.genabel.org >> >> Anatoly, can you please take care of this some time this week, if >> possible? I presume we want to redirect traffic from >> >> http://mga.bionet.nsc.ru/nlru/GenABEL/ >> http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ >> http://mga.bionet.nsc.ru/~yurii/ABEL/ > I changed index.html file on this pages with > > It works. > >> >> and (ideally) all other content in these directories. This is >> important: e.g. if you search google for "ABEL tutorial" you get >> >> http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ABEL-tutorial.pdf > The page is redirected too. But there is no pdf-file > http://www.genabel.org/old/pdfs/ABEL-tutorial.pdf > > Which pages should be redirected else? > > Do you like this way of redirection? > >> >> as the top hit. >> >> Just to remind what we had in our discussion about re-direction: >> >> "Good" procedure: >> >> >> >> >> This depends a bit on the level of access that you have on MGA. IIRC >> >> you >> >> have the files in your ~/public_html dir, right? There are a couple of >> >> ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite >> >> enabled we can use that. See e.g. >> >> http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an >> >> example. >> >> >> > >> >> Simple procedure (but we will be penalized by search engines): >> >> > something more simple, which I used for old-old-old >> > page of GenABEL, see >> > >> > http://mga.bionet.nsc.ru/nlru/GenABEL >> > >> > basically, it boils down to this HTML line >> > >> > > > Anatoly > > -- Sent from my mobile device From kianvl at mail.ru Thu Mar 17 13:29:17 2011 From: kianvl at mail.ru (Anatoliy Kirichenko) Date: Thu, 17 Mar 2011 15:29:17 +0300 Subject: [GenABEL-dev] =?utf-8?q?web-server_todo=27s?= In-Reply-To: References: Message-ID: Dear Yurii, dear All, > Anatoly, can you please take care of this some time this week, if > possible? I presume we want to redirect traffic from > > http://mga.bionet.nsc.ru/nlru/GenABEL/ > http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ > http://mga.bionet.nsc.ru/~yurii/ABEL/ > > and (ideally) all other content in these directories. This is > important: e.g. if you search google for "ABEL tutorial" you get > > http://mga.bionet.nsc.ru/~yurii/ABEL/GenABEL/ABEL-tutorial.pdf > > as the top hit. > > Just to remind what we had in our discussion about re-direction: > > "Good" procedure: > > >> This depends a bit on the level of access that you have on MGA. IIRC you > >> have the files in your ~/public_html dir, right? There are a couple of > >> ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite > >> enabled we can use that. See e.g. > >> http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an > >> example. Everything was done in this way! :) > > Simple procedure (but we will be penalized by search engines): > > > something more simple, which I used for old-old-old > > page of GenABEL, see > > > > http://mga.bionet.nsc.ru/nlru/GenABEL > > > > basically, it boils down to this HTML line > > > > From yurii.aulchenko at gmail.com Thu Mar 17 14:08:39 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 17 Mar 2011 14:08:39 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References:

Message-ID: I believe that technically our transition to www.genabel.org web-site is pretty much complete. Thanks a lot, Anatoly, and many thanks to Lennart for making this happening! We still need to refine the content, but I think this is something which will happen naturally through the devel- and user- community response. Yurii On Thu, Mar 17, 2011 at 1:29 PM, Anatoliy Kirichenko wrote: > Dear Yurii, dear All, > >> >> ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite >> >> enabled we can use that. See e.g. >> >> http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an >> >> example. > > Everything was done in this way! :) > From l.karssen at erasmusmc.nl Thu Mar 17 14:15:29 2011 From: l.karssen at erasmusmc.nl (L.C. Karssen) Date: Thu, 17 Mar 2011 14:15:29 +0100 Subject: [GenABEL-dev] web-server todo's In-Reply-To: References:

Message-ID: <1300367729.18534.12.camel@rubidium> On Thu, 2011-03-17 at 14:08 +0100, Yurii Aulchenko wrote: > I believe that technically our transition to www.genabel.org web-site > is pretty much complete. Don't forget that we should also point the DNS records for the genabel.net and genabel.com domains to their .org counterparts. Lennart. > > Thanks a lot, Anatoly, and many thanks to Lennart for making this happening! > > We still need to refine the content, but I think this is something > which will happen naturally through the devel- and user- community > response. > > Yurii > > On Thu, Mar 17, 2011 at 1:29 PM, Anatoliy Kirichenko wrote: > > Dear Yurii, dear All, > > > >> >> ways to redirect pages. If the sysadmin of MGA has Apache's mod_rewrite > >> >> enabled we can use that. See e.g. > >> >> http://www.stevenhargrove.com/redirect-web-pages/#301modrewrite for an > >> >> example. > > > > Everything was done in this way! :) > > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel -- ----------------------------------------------- L.C. Karssen Erasmus MC Department of Epidemiology Room Ee2224 Postbus 2040 3000 CA Rotterdam The Netherlands phone: +31-10-7044217 fax: +31-10-7044657 email: l.karssen at erasmusmc.nl GPG key ID: 0E1D39E3 ----------------------------------------------- -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From yurii.aulchenko at gmail.com Tue Mar 22 02:27:09 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 22 Mar 2011 02:27:09 +0100 Subject: [GenABEL-dev] [Genabel-commits] r701 - in pkg/GenABEL: . R inst/unitTests man In-Reply-To: <20110322011758.0F0FC9CF91@r-forge.r-project.org> References: <20110322011758.0F0FC9CF91@r-forge.r-project.org> Message-ID: Dear All, In this commit, I have added 'findRelatives' and four supporting functions. This function make inferences on relationship from genomic data. Compared to guessing relations from genomic kinship matrix, this procedure offers several enhancements: (1) by use of IBD/IBS 3-state space, it allows to distinguish between some pairs, which have the same kinship (e.g. parent-offspring from brother-sister; uncle-nephew from grandparent-grandchild, etc.) (2) it reports likelihood, allowing for more rigorous inferences. All functions are documented using 'roxygen'; RUnit test designed for 'findRelatives'. More details at http://lists.r-forge.r-project.org/pipermail/genabel-commits/2011-March/000084.html best wishes, Yurii On Tue, Mar 22, 2011 at 2:17 AM, wrote: > Author: yurii > Date: 2011-03-22 02:17:57 +0100 (Tue, 22 Mar 2011) > New Revision: 701 > > Added: > ? pkg/GenABEL/R/blurGenotype.R > ? pkg/GenABEL/R/findRelatives.R > ? pkg/GenABEL/R/generateOffspring.R > ? pkg/GenABEL/R/getLogLikelihoodGivenRelation.R > ? pkg/GenABEL/R/makeTransitionMatrix.R > ? pkg/GenABEL/inst/unitTests/runit.findRelatives.R > ? pkg/GenABEL/man/blurGenotype.Rd > ? pkg/GenABEL/man/findRelatives.Rd > ? pkg/GenABEL/man/generateOffspring.Rd > ? pkg/GenABEL/man/getLogLikelihoodGivenRelation.Rd > ? pkg/GenABEL/man/makeTransitionMatrix.Rd > Modified: > ? pkg/GenABEL/CHANGES.LOG > ? pkg/GenABEL/NAMESPACE > ? pkg/GenABEL/generate_documentation.R > ? pkg/GenABEL/inst/unitTests/report.html > ? pkg/GenABEL/inst/unitTests/report.txt > ? pkg/GenABEL/inst/unitTests/reportSummary.txt > ? pkg/GenABEL/man/GenABEL-package.Rd > ? pkg/GenABEL/man/polygenic.Rd > ? pkg/GenABEL/man/polygenic_hglm.Rd > Log: > added 'findRelatives' and four supporting functions; also some accumulated changes > From xia.shen at lcb.uu.se Tue Mar 22 17:33:11 2011 From: xia.shen at lcb.uu.se (Xia Shen) Date: Tue, 22 Mar 2011 17:33:11 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se>

<3CF9915B-8747-405B-B222-846EEF3F9471@lcb.uu.se> Message-ID: <07250A8D-694C-4209-A6EB-33FD57DAA505@lcb.uu.se> Hi Yurii, Attached is an updated version of 'polygenic_hglm' function with documentation. I've fixed: - Brief description of the estimation procedure; - Reference list; - An example for testing significance of the polygenic variance component; - Some minor typos. Perhaps the default 'conv' can be relaxed a bit. 1e-6 is good enough to me, and also 1e-8 might be two strict for some binary trait - getting a precision of 8 digits for binary data is too difficult, sometimes it might not even converge. We are working on updating the 'hglm' package on CRAN, where I will make extracting internal variance-covariance matrix possible. Cheers Xia -------------- next part -------------- A non-text attachment was scrubbed... Name: polygenic_hglm.R Type: application/octet-stream Size: 4993 bytes Desc: not available URL: -------------- next part -------------- On Mar 7, 2011, at 10:10 PM, Yurii Aulchenko wrote: > Xia et al., > > here are results > >> I just 'sudo-randomly' tried some different argument value in the hglm function call, and happened to have solved the bias issue (at least it seems to be so). Please try putting the following arguments in your hglm wrapper: >> >> hglm(y = ph, X = desmat, Z = L, conv = 1e-6, maxit = 40) >> >> I re-did 100 simulations and the bias is not sig from 0 with p-value = 0.77! >> > > To be even more strict, I have set > > hglm( ... , conv = 1e-8, maxit = 100) > > and ran the tests. Results attached as PDF (+ the script) > > First, some great news: > > See figure, columns a) and b): there is almost 1-to-1 identity in h2 > estimates generated by polygenic_hglm and FMM. I take this as a sign > that both of them do a great job, while 'old' polygenic underestimates > h2 slightly. > > Do not quite understand whathappens with old -- and not quite willing > to dig into it yet, as lambdas are closest to 1 with old -- so I think > we stick to the old form of old-polygenic till we can explain the > whole story. Will wait till polygenic-FMM is out to do another round > of comparisons. > > So, I thin Xia's suggestion > >> I guess the problem here is just numerical. Since IBS matrix is generally sparse (close to identity), the estimation of variance components will be sensitive numerically. Just make the convergence criteria more strict, and it works. >> > > is right. And the 0.4-strangeness must be something to deal with the > starting values, what do you think? > >> Note that this might slow down hglm a bit (not very much I guess), but I think it should be more than 20x faster than the old polygenic. Not familiar with fmm, but: 1. our hglm can deal with any GLM family of the trait distribution, even non-normal genetic effects. 2. we produce standard errors for everything - every part is checkable using h-likelihood :) >> > > It did become slower with the strict criteria -- now speed-up is x5.3 > (range: x1.7 - x80.8). Still, this is quite incredible :) Next step > will be William's polygenic_fmm -- his speed-up is > more-than-incredible :) > > Ok, what about finalizing this so we have 'polygenic_hglm' > ready-to-use in the next release? > > Here is a plan: > > I will take care of updating documentation to 'polygenic' so it > references to 'polygenic_hglm' as an alternative; also main package > documentation. You will see that on genabel-commits. > > ---- Things we absolutely need before it can be released: > > * Test if resulting object works at all with grammar procedure and > produces roughly similar results with old-polygenic object > * Pull the se's of the estimates in the output -- this allows for > testing significance of covariates > * Update documentation: > ** Xia, please make sure you mention everything you want to mention -- > give proper credit, references, etc. > ** Better explanation of the method. We can of cause always refer > people to the main 'hglm' man page, but it will be nice to outline -- > e.g. that REML is used by default > ** Examples: how do you test for a SNP effect? See for example of > situation with old polygenic: > > http://forum.genabel.org/viewtopic.php?f=6&t=134&sid=c5dcf7fc39fd3f7d07b1308bafed5286 > > > ----- Things, which will be great to have, but not critical to the release: > > ** Similar to old-polygenic fixh2 argument -- in which case we can > also output -2xLogLik (but need to make sure people understand that eg > in REML setting this can only be used for testing random effect / h2) > [also look up what are other options available in old-polygenic but > not in polygenic_hglm] > ** optimization of the code -- many things which are computed in > wrapper should have been computed by 'hglm', so we can gain some speed > > ---- Nice, but not-so-urgent things: > > * unify and formalize 'polygenic' as S4-class > > best wishes, and hear from you, > Yurii > > PS To be followed-up later: why the hell does old-polygenic+mmscore > still generates Lambdas closest to 1 (see figure, columns c), d))? > Something to deal with uncertainty in gkin, as suggested by William > (very interesting question!), conservativity of the 'mmscore' as > suggested by Yurii, etc. At the end, we want to know which one is the > most powerful procedure. This is something we can test directly. > From yurii.aulchenko at gmail.com Tue Mar 22 20:43:44 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Tue, 22 Mar 2011 20:43:44 +0100 Subject: [GenABEL-dev] faster polygenic In-Reply-To: <07250A8D-694C-4209-A6EB-33FD57DAA505@lcb.uu.se> References:

<49EAF07F-6BCD-43CA-8A8E-2513CB2517E8@lcb.uu.se>

<3CF9915B-8747-405B-B222-846EEF3F9471@lcb.uu.se> <07250A8D-694C-4209-A6EB-33FD57DAA505@lcb.uu.se> Message-ID: Hi Xia, Many thanks! The code looks good to me, and examples are very nice! > - Reference list; > I suggest we change order to chronological, so your paper is 1st, then Lee2001, Lee1996 > Perhaps the default 'conv' can be relaxed a bit. 1e-6 is good enough to me, and also 1e-8 might be two strict for some binary trait - getting a precision of 8 digits for binary data is too difficult, sometimes it might not even converge. > I agree. This will probably make it much faster. But we should tell that in the documentation -- like "if you want really great convergence, use 1e-8..." > We are working on updating the 'hglm' package on CRAN, where I will make extracting internal variance-covariance matrix possible. > These are great news! One thing which can make polygenic_hglm even more apparent news: I understood that for the fixed effects, you get standard errors of the estimates, correct? And these can be used for significance testing, right? E.g. SNP or haplotype significance (if one puts a SNP as covar). This would be extremely valuable, see discussion http://forum.genabel.org/viewtopic.php?f=6&t=134 If easy, I suggest to pull fixed effects SE more close to the user, and provide an example of doing test for a SNP. I really think this will help polygenic_hglm to make to the news-lines -- and this is something quite unique [not in 'polygenic', not, to my knowledge, for FMM] :) Yurii From pirastu at burlo.trieste.it Fri Mar 25 13:02:20 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Fri, 25 Mar 2011 13:02:20 +0100 (CET) Subject: [GenABEL-dev] descriptives.trait Patch Message-ID: <49674.193.62.206.85.1301054540.squirrel@webmail-rupar.regione.fvg.it> Dear all, attached you can find a patch to fix the bugs described underneath. To install from the root directory of /pkg: patch -p0 < patch_descriptives.trait. Best Nicola Patch for bugs #1184 #1185 #1259 Bug [#1184] descriptives.trait can not see the variable in by.var argument unless full path specified Description: https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1184&group_id=505&atid=2058 Changes: now by.var can be specified aslo as :"var_name" or ~var_name Test: > library(GenABEL) > data(ge03d2ex) > descriptives.trait(ge03d2ex,by.var="sex") No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw id 64 NA NA 72 NA NA NA NA sex 64 NA NA 72 NA NA NA NA age 64 46.942 12.479 72 50.959 13.107 0.070 0.081 dm2 64 0.547 0.502 72 0.708 0.458 0.053 0.052 height 64 162.680 6.819 71 175.534 7.943 0.000 0.000 weight 64 78.605 26.908 71 95.322 21.441 0.000 0.000 diet 64 0.109 0.315 72 0.014 0.118 0.025 0.019 bmi 64 29.604 9.506 71 30.930 6.547 0.352 0.040 Pexact id NA sex NA age NA dm2 0.074 height NA weight NA diet 0.026 bmi NA > descriptives.trait(ge03d2ex,by.var=~sex) No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw id 64 NA NA 72 NA NA NA NA sex 64 NA NA 72 NA NA NA NA age 64 46.942 12.479 72 50.959 13.107 0.070 0.081 dm2 64 0.547 0.502 72 0.708 0.458 0.053 0.052 height 64 162.680 6.819 71 175.534 7.943 0.000 0.000 weight 64 78.605 26.908 71 95.322 21.441 0.000 0.000 diet 64 0.109 0.315 72 0.014 0.118 0.025 0.019 bmi 64 29.604 9.506 71 30.930 6.547 0.352 0.040 Pexact id NA sex NA age NA dm2 0.074 height NA weight NA diet 0.026 bmi NA Bug [#1185] descriptives.trait with missings in by.var argument Description: https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1185&group_id=505&atid=2058 Fixed: now by.var argument acceptes missing Test: > phdata(ge03d2ex)$sex[2] <- NA > descriptives.trait(ge03d2ex,by.var=phdata(ge03d2ex)$sex) No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw id 63 NA NA 72 NA NA NA NA sex 63 NA NA 72 NA NA NA NA age 63 47.003 12.570 72 50.959 13.107 0.076 0.088 dm2 63 0.540 0.502 72 0.708 0.458 0.045 0.044 height 63 162.573 6.819 71 175.534 7.943 0.000 0.000 weight 63 77.651 26.009 71 95.322 21.441 0.000 0.000 diet 63 0.111 0.317 72 0.014 0.118 0.024 0.017 bmi 63 29.306 9.277 71 30.930 6.547 0.250 0.025 Pexact id NA sex NA age NA dm2 0.051 height NA weight NA diet 0.025 bmi NA Bug [#1259] descriptives.trait with by.var argument does not work (contributed by Richard Ahn) Description: https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1259&group_id=505&atid=2058 Comments: The problem was that for one of the traits one of the categories had all missing data, this made descriptives.trait crush. Fixed: descriptives.trait does not crush anymore. Test: nicola at nicola-VirtualBox:~/Scrivania$ R -f test.R WARNING: ignoring environment value of R_HOME R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R ? un software libero ed ? rilasciato SENZA ALCUNA GARANZIA. Siamo ben lieti se potrai redistribuirlo, ma sotto certe condizioni. Scrivi 'license()' o 'licence()' per dettagli su come distribuirlo. R ? un progetto di collaborazione con molti contributi esterni. Scrivi 'contributors()' per maggiori informazioni e 'citation()' per sapere come citare R o i pacchetti di R nelle pubblicazioni. Scrivi 'demo()' per una dimostrazione, 'help()' per la guida in linea, o 'help.start()' per l'help navigabile con browser HTML. Scrivi 'q()' per uscire da R. > library(GenABEL) Carico il pacchetto richiesto: MASS GenABEL v. 1.6-6 (February 24, 2011) loaded Installed GenABEL version (1.6-6) is not the same as stable version available from CRAN (1.6-5). Unless used intentionally, consider updating to the latest CRAN version. For that, use 'install.packages("GenABEL")', or ask your system administrator to update the package. > convert.snp.illumina(infile="test_markers", outfile="test.raw") Reading genotypes from file 'test_markers' ... Writing to file 'test.raw' ... ... done. > test = load.gwaa.data(pheno="test_phenos", geno="test.raw", force=T) ids loaded... marker names loaded... chromosome data loaded... map data loaded... allele coding data loaded... strand data loaded... genotype data loaded... snp.data object created... assignment of gwaa.data object FORCED; X-errors were not checked! > attach(phdata(test)) > descriptives.trait(data=test, by = bt) Error in t.test.formula(ctrao ~ svar) : grouping factor must have exactly 2 levels Error in kruskal.test.default(c(6L, 32L, 35L, 33L, 52L, 25L), c(1L, 1L, : all observations are in the same group No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw Pexact id 5 NA NA 6 NA NA NA NA NA sex 5 0.2 0.447 6 0.333 0.516 0.657 0.637 1 age 0 NaN NA 6 30.500 14.977 NA NA NA bt 5 NA NA 6 NA NA NA NA NA > > From pirastu at burlo.trieste.it Fri Mar 25 13:03:57 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Fri, 25 Mar 2011 13:03:57 +0100 (CET) Subject: [GenABEL-dev] [Fwd: descriptives.trait Patch] Message-ID: <49692.193.62.206.85.1301054637.squirrel@webmail-rupar.regione.fvg.it> Sorry I forgot the attachment. Dear all, attached you can find a patch to fix the bugs described underneath. To install from the root directory of /pkg: patch -p0 < patch_descriptives.trait. Best Nicola Patch for bugs #1184 #1185 #1259 Bug [#1184] descriptives.trait can not see the variable in by.var argument unless full path specified Description: https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1184&group_id=505&atid=2058 Changes: now by.var can be specified aslo as :"var_name" or ~var_name Test: > library(GenABEL) > data(ge03d2ex) > descriptives.trait(ge03d2ex,by.var="sex") No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw id 64 NA NA 72 NA NA NA NA sex 64 NA NA 72 NA NA NA NA age 64 46.942 12.479 72 50.959 13.107 0.070 0.081 dm2 64 0.547 0.502 72 0.708 0.458 0.053 0.052 height 64 162.680 6.819 71 175.534 7.943 0.000 0.000 weight 64 78.605 26.908 71 95.322 21.441 0.000 0.000 diet 64 0.109 0.315 72 0.014 0.118 0.025 0.019 bmi 64 29.604 9.506 71 30.930 6.547 0.352 0.040 Pexact id NA sex NA age NA dm2 0.074 height NA weight NA diet 0.026 bmi NA > descriptives.trait(ge03d2ex,by.var=~sex) No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw id 64 NA NA 72 NA NA NA NA sex 64 NA NA 72 NA NA NA NA age 64 46.942 12.479 72 50.959 13.107 0.070 0.081 dm2 64 0.547 0.502 72 0.708 0.458 0.053 0.052 height 64 162.680 6.819 71 175.534 7.943 0.000 0.000 weight 64 78.605 26.908 71 95.322 21.441 0.000 0.000 diet 64 0.109 0.315 72 0.014 0.118 0.025 0.019 bmi 64 29.604 9.506 71 30.930 6.547 0.352 0.040 Pexact id NA sex NA age NA dm2 0.074 height NA weight NA diet 0.026 bmi NA Bug [#1185] descriptives.trait with missings in by.var argument Description: https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1185&group_id=505&atid=2058 Fixed: now by.var argument acceptes missing Test: > phdata(ge03d2ex)$sex[2] <- NA > descriptives.trait(ge03d2ex,by.var=phdata(ge03d2ex)$sex) No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw id 63 NA NA 72 NA NA NA NA sex 63 NA NA 72 NA NA NA NA age 63 47.003 12.570 72 50.959 13.107 0.076 0.088 dm2 63 0.540 0.502 72 0.708 0.458 0.045 0.044 height 63 162.573 6.819 71 175.534 7.943 0.000 0.000 weight 63 77.651 26.009 71 95.322 21.441 0.000 0.000 diet 63 0.111 0.317 72 0.014 0.118 0.024 0.017 bmi 63 29.306 9.277 71 30.930 6.547 0.250 0.025 Pexact id NA sex NA age NA dm2 0.051 height NA weight NA diet 0.025 bmi NA Bug [#1259] descriptives.trait with by.var argument does not work (contributed by Richard Ahn) Description: https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1259&group_id=505&atid=2058 Comments: The problem was that for one of the traits one of the categories had all missing data, this made descriptives.trait crush. Fixed: descriptives.trait does not crush anymore. Test: nicola at nicola-VirtualBox:~/Scrivania$ R -f test.R WARNING: ignoring environment value of R_HOME R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R ? un software libero ed ? rilasciato SENZA ALCUNA GARANZIA. Siamo ben lieti se potrai redistribuirlo, ma sotto certe condizioni. Scrivi 'license()' o 'licence()' per dettagli su come distribuirlo. R ? un progetto di collaborazione con molti contributi esterni. Scrivi 'contributors()' per maggiori informazioni e 'citation()' per sapere come citare R o i pacchetti di R nelle pubblicazioni. Scrivi 'demo()' per una dimostrazione, 'help()' per la guida in linea, o 'help.start()' per l'help navigabile con browser HTML. Scrivi 'q()' per uscire da R. > library(GenABEL) Carico il pacchetto richiesto: MASS GenABEL v. 1.6-6 (February 24, 2011) loaded Installed GenABEL version (1.6-6) is not the same as stable version available from CRAN (1.6-5). Unless used intentionally, consider updating to the latest CRAN version. For that, use 'install.packages("GenABEL")', or ask your system administrator to update the package. > convert.snp.illumina(infile="test_markers", outfile="test.raw") Reading genotypes from file 'test_markers' ... Writing to file 'test.raw' ... ... done. > test = load.gwaa.data(pheno="test_phenos", geno="test.raw", force=T) ids loaded... marker names loaded... chromosome data loaded... map data loaded... allele coding data loaded... strand data loaded... genotype data loaded... snp.data object created... assignment of gwaa.data object FORCED; X-errors were not checked! > attach(phdata(test)) > descriptives.trait(data=test, by = bt) Error in t.test.formula(ctrao ~ svar) : grouping factor must have exactly 2 levels Error in kruskal.test.default(c(6L, 32L, 35L, 33L, 52L, 25L), c(1L, 1L, : all observations are in the same group No(by.var=0) Mean SD No(by.var=1) Mean SD Ptt Pkw Pexact id 5 NA NA 6 NA NA NA NA NA sex 5 0.2 0.447 6 0.333 0.516 0.657 0.637 1 age 0 NaN NA 6 30.500 14.977 NA NA NA bt 5 NA NA 6 NA NA NA NA NA > > -------------- next part -------------- A non-text attachment was scrubbed... Name: patch_descriptives.trait Type: application/octet-stream Size: 5753 bytes Desc: not available URL: From lennart at karssen.org Sun Mar 27 13:12:32 2011 From: lennart at karssen.org (L.C. Karssen) Date: Sun, 27 Mar 2011 13:12:32 +0200 Subject: [GenABEL-dev] void error(char) function definition, where to find it? Message-ID: <1301224352.2781.16.camel@barabas> Hi all, I've been trying to fix ProbABEL bug #1339 [1] (problem that ProbABEL doesn't compile because frerror.h is missing). This missing header file (and the accompanying .cpp file) define a function void error(const char * format, ... ); The frerror files have been deleted from the fvlib directory in svn in r599, but grep-ing the complete genabel/pkg directory I still see that several parts of the code call a function error(char), for example Yurii's convert_snp_merlin.cpp (GenABEL) and Maksim's supplementary_functions.cpp (VariABEL). Since I couldn't find a definition of the function in the code of the GenABEL suite, I assume it is defined elsewhere. I've tried to find out where this function is defined, but I failed. Who can point me to the (header) file where the error() function is defined? Thanks, Lennart. [1] https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1339&group_id=505&atid=2058 -- ----------------------------------------------------------------- L.C. Karssen Utrecht The Netherlands lennart at karssen.org http://blog.karssen.org Stuur me aub geen Word of Powerpoint bestanden! Zie http://www.gnu.org/philosophy/no-word-attachments.nl.html ------------------------------------------------------------------ -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part URL: From mgmaria at yandex.ru Tue Mar 29 00:50:38 2011 From: mgmaria at yandex.ru (Maria G) Date: Tue, 29 Mar 2011 02:50:38 +0400 Subject: [GenABEL-dev] Journal Club on Statistical Genomics Message-ID: <723171301352638@web124.yandex.ru> Dear all, My greetings to all of yours! Concerning the content of the www.genabel.org, what would you say about the opening a new forum with the title something like: ?Journal Club on Statistical Genomics? ?! It might be useful for both experienced and beginners in statistical background of GenABEL applications. The general idea is to give us a possibility to share the knowledge on biostatistics based on the papers available in Internet. Please, find the brief description of the forum attached in the file ?Journal_Club_on_Statistical_Genomics.rtf?. As I would see that, in the Forum there might be at least 3 separate sections: (1) Bright papers: ?Good to know ?? -> To discuss any interesting paper in the frame of Statistical Genomics (2) Essential notions: ?What is ? ?? -> To clarify any basic concepts of Statistical Genomics (3) Methodological nuances: ?How to? ?? -> To focus on different methods of Statistical Genomics If it could be of interest I would definitely participate in the forum organization Best wishes Masha -------------- next part -------------- A non-text attachment was scrubbed... Name: Journal_Club_on_Statistical_Genomics.rtf Type: application/rtf Size: 90268 bytes Desc: not available URL: From yurii.aulchenko at gmail.com Thu Mar 31 00:15:52 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 31 Mar 2011 00:15:52 +0200 Subject: [GenABEL-dev] descriptives.trait Patch In-Reply-To: <49674.193.62.206.85.1301054540.squirrel@webmail-rupar.regione.fvg.it> References: <49674.193.62.206.85.1301054540.squirrel@webmail-rupar.regione.fvg.it> Message-ID: Dear Nicola, Many thanks for submitting this patch! These little problems were pestering 'descriptive.traits' from the very moment the procedure was invented -- so it is great news indeed and a great relief to have these problems fixed! I see no problem with applying this patch to the main release (and within 5 days there were no concerns from others). So, I will apply this patch + add RUnit regression tests for these three bugs -- and when the commit (or any problems) will be there you will hear about this on the mailing list. best wishes, Yurii On Fri, Mar 25, 2011 at 1:02 PM, wrote: > Dear all, > > attached you can find a patch to fix the bugs described underneath. To > install from the root directory of /pkg: patch -p0 < > patch_descriptives.trait. > > Best > > Nicola > > Patch for bugs #1184 #1185 #1259 > > Bug [#1184] descriptives.trait can not see the variable in by.var argument > unless full path specified > > Description: > https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1184&group_id=505&atid=2058 > > Changes: now by.var can be specified aslo as :"var_name" or ~var_name > > Test: > >> library(GenABEL) >> data(ge03d2ex) >> descriptives.trait(ge03d2ex,by.var="sex") > ? ? ? No(by.var=0) ? ?Mean ? ? SD No(by.var=1) ? ?Mean ? ? SD ? Ptt ? Pkw > id ? ? ? ? ? ? ? 64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? ?NA > sex ? ? ? ? ? ? ?64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? ?NA > age ? ? ? ? ? ? ?64 ?46.942 12.479 ? ? ? ? ? 72 ?50.959 13.107 0.070 0.081 > dm2 ? ? ? ? ? ? ?64 ? 0.547 ?0.502 ? ? ? ? ? 72 ? 0.708 ?0.458 0.053 0.052 > height ? ? ? ? ? 64 162.680 ?6.819 ? ? ? ? ? 71 175.534 ?7.943 0.000 0.000 > weight ? ? ? ? ? 64 ?78.605 26.908 ? ? ? ? ? 71 ?95.322 21.441 0.000 0.000 > diet ? ? ? ? ? ? 64 ? 0.109 ?0.315 ? ? ? ? ? 72 ? 0.014 ?0.118 0.025 0.019 > bmi ? ? ? ? ? ? ?64 ?29.604 ?9.506 ? ? ? ? ? 71 ?30.930 ?6.547 0.352 0.040 > ? ? ? Pexact > id ? ? ? ? NA > sex ? ? ? ?NA > age ? ? ? ?NA > dm2 ? ? 0.074 > height ? ? NA > weight ? ? NA > diet ? ?0.026 > bmi ? ? ? ?NA >> descriptives.trait(ge03d2ex,by.var=~sex) > ? ? ? No(by.var=0) ? ?Mean ? ? SD No(by.var=1) ? ?Mean ? ? SD ? Ptt ? Pkw > id ? ? ? ? ? ? ? 64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? ?NA > sex ? ? ? ? ? ? ?64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? ?NA > age ? ? ? ? ? ? ?64 ?46.942 12.479 ? ? ? ? ? 72 ?50.959 13.107 0.070 0.081 > dm2 ? ? ? ? ? ? ?64 ? 0.547 ?0.502 ? ? ? ? ? 72 ? 0.708 ?0.458 0.053 0.052 > height ? ? ? ? ? 64 162.680 ?6.819 ? ? ? ? ? 71 175.534 ?7.943 0.000 0.000 > weight ? ? ? ? ? 64 ?78.605 26.908 ? ? ? ? ? 71 ?95.322 21.441 0.000 0.000 > diet ? ? ? ? ? ? 64 ? 0.109 ?0.315 ? ? ? ? ? 72 ? 0.014 ?0.118 0.025 0.019 > bmi ? ? ? ? ? ? ?64 ?29.604 ?9.506 ? ? ? ? ? 71 ?30.930 ?6.547 0.352 0.040 > ? ? ? Pexact > id ? ? ? ? NA > sex ? ? ? ?NA > age ? ? ? ?NA > dm2 ? ? 0.074 > height ? ? NA > weight ? ? NA > diet ? ?0.026 > bmi ? ? ? ?NA > > Bug [#1185] descriptives.trait with missings in by.var argument > > Description: > https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1185&group_id=505&atid=2058 > > Fixed: now by.var argument acceptes missing > > Test: > >> phdata(ge03d2ex)$sex[2] <- NA >> descriptives.trait(ge03d2ex,by.var=phdata(ge03d2ex)$sex) > ? ? ? No(by.var=0) ? ?Mean ? ? SD No(by.var=1) ? ?Mean ? ? SD ? Ptt ? Pkw > id ? ? ? ? ? ? ? 63 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? ?NA > sex ? ? ? ? ? ? ?63 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? ?NA > age ? ? ? ? ? ? ?63 ?47.003 12.570 ? ? ? ? ? 72 ?50.959 13.107 0.076 0.088 > dm2 ? ? ? ? ? ? ?63 ? 0.540 ?0.502 ? ? ? ? ? 72 ? 0.708 ?0.458 0.045 0.044 > height ? ? ? ? ? 63 162.573 ?6.819 ? ? ? ? ? 71 175.534 ?7.943 0.000 0.000 > weight ? ? ? ? ? 63 ?77.651 26.009 ? ? ? ? ? 71 ?95.322 21.441 0.000 0.000 > diet ? ? ? ? ? ? 63 ? 0.111 ?0.317 ? ? ? ? ? 72 ? 0.014 ?0.118 0.024 0.017 > bmi ? ? ? ? ? ? ?63 ?29.306 ?9.277 ? ? ? ? ? 71 ?30.930 ?6.547 0.250 0.025 > ? ? ? Pexact > id ? ? ? ? NA > sex ? ? ? ?NA > age ? ? ? ?NA > dm2 ? ? 0.051 > height ? ? NA > weight ? ? NA > diet ? ?0.025 > bmi ? ? ? ?NA > > Bug [#1259] descriptives.trait with by.var argument does not work > (contributed by Richard Ahn) > > Description: > https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1259&group_id=505&atid=2058 > > Comments: The problem was that for one of the traits one of the categories > had all missing data, this made descriptives.trait crush. > > Fixed: descriptives.trait does not crush anymore. > > Test: > > nicola at nicola-VirtualBox:~/Scrivania$ R -f test.R > WARNING: ignoring environment value of R_HOME > > R version 2.11.1 (2010-05-31) > Copyright (C) 2010 The R Foundation for Statistical Computing > ISBN 3-900051-07-0 > > R ? un software libero ed ? rilasciato SENZA ALCUNA GARANZIA. > Siamo ben lieti se potrai redistribuirlo, ma sotto certe condizioni. > Scrivi 'license()' o 'licence()' per dettagli su come distribuirlo. > > R ? un progetto di collaborazione con molti contributi esterni. > Scrivi 'contributors()' per maggiori informazioni e 'citation()' > per sapere come citare R o i pacchetti di R nelle pubblicazioni. > > Scrivi 'demo()' per una dimostrazione, 'help()' per la guida in linea, o > 'help.start()' per l'help navigabile con browser HTML. > Scrivi 'q()' per uscire da R. > >> library(GenABEL) > Carico il pacchetto richiesto: MASS > GenABEL v. 1.6-6 (February 24, 2011) loaded > > Installed GenABEL version (1.6-6) is not the same as stable > version available from CRAN (1.6-5). Unless used intentionally, > consider updating to the latest CRAN version. For that, use > 'install.packages("GenABEL")', or ask your system administrator > to update the package. > >> convert.snp.illumina(infile="test_markers", outfile="test.raw") > Reading genotypes from file 'test_markers' ... > Writing to file 'test.raw' ... > ... done. >> test = load.gwaa.data(pheno="test_phenos", geno="test.raw", force=T) > ids loaded... > marker names loaded... > chromosome data loaded... > map data loaded... > allele coding data loaded... > strand data loaded... > genotype data loaded... > snp.data object created... > assignment of gwaa.data object FORCED; X-errors were not checked! >> attach(phdata(test)) >> descriptives.trait(data=test, by = bt) > Error in t.test.formula(ctrao ~ svar) : > ?grouping factor must have exactly 2 levels > Error in kruskal.test.default(c(6L, 32L, 35L, 33L, 52L, 25L), c(1L, 1L, ?: > ?all observations are in the same group > ? ?No(by.var=0) Mean ? ?SD No(by.var=1) ? Mean ? ? SD ? Ptt ? Pkw Pexact > id ? ? ? ? ? ? 5 ? NA ? ?NA ? ? ? ? ? ?6 ? ? NA ? ? NA ? ?NA ? ?NA ? ? NA > sex ? ? ? ? ? ?5 ?0.2 0.447 ? ? ? ? ? ?6 ?0.333 ?0.516 0.657 0.637 ? ? ?1 > age ? ? ? ? ? ?0 ?NaN ? ?NA ? ? ? ? ? ?6 30.500 14.977 ? ?NA ? ?NA ? ? NA > bt ? ? ? ? ? ? 5 ? NA ? ?NA ? ? ? ? ? ?6 ? ? NA ? ? NA ? ?NA ? ?NA ? ? NA >> >> > > > > > > > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > From yurii.aulchenko at gmail.com Thu Mar 31 01:53:26 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 31 Mar 2011 01:53:26 +0200 Subject: [GenABEL-dev] Journal Club on Statistical Genomics In-Reply-To: <723171301352638@web124.yandex.ru> References: <723171301352638@web124.yandex.ru> Message-ID: Dear Masha, > Concerning the content of the www.genabel.org, what would you say about the opening a new forum with the title something like: ??Journal Club on Statistical Genomics? ?! > > It might be useful for both experienced and beginners in statistical background of GenABEL applications. The general idea is to give us a possibility to share the knowledge on biostatistics based on the papers available in Internet. Many thanks for this suggestion! I am personally very much in favor of doing this. This would be in a good agreement with open-source-based methodology development framework we are trying to build in GenABEL-project; such forum may lower 'entry threshold' (compared to semi-closed genabel-devel mailing list) and will extend methodology discussion to a wider audience. Taking GenABEL-project-centric view, I feel that having such forum is likely to bring a broader range of methodological topics into the project's framework, and potentially attract more contributors. >Please, find the brief description of the forum attached in the file ?Journal_Club_on_Statistical_Genomics.rtf?. As I would see that, in the Forum there might be at least 3 separate sections: > > (1) ? ? Bright papers: ?Good to know ?? ?-> ?To discuss any interesting paper in the frame of Statistical Genomics > (2) ? ? Essential notions: ?What is ? ?? -> To clarify any basic concepts of Statistical Genomics > (3) ? ? Methodological nuances: ?How to? ?? -> To focus on different methods of Statistical Genomics > I am totally supportive of (1) and quite positive about (2). As for (3), I wonder how much it may overlap with (2) and the mainstream GenABEL-user forum. Would you have a clear-cut question of type (3), which does not fall into (2) and GenABEL-user? > If it could be of interest I would definitely participate in the forum organization > Let us discuss your suggestion at this mailing list, and, if positive, we would definitely be more than willing to provide infrastructure for such forum. Given the agreement, and the fact that you are already moderating GenABEL-user forum, there is no doubt that you are capable of managing such a forum, and you willingness to invent it will be very much appreciated! To conclude: my vote is "yes" (with some reservations) with best wishes, Yurii From yurii.aulchenko at gmail.com Thu Mar 31 02:00:24 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 31 Mar 2011 02:00:24 +0200 Subject: [GenABEL-dev] [Genabel-commits] r704 - in pkg/GenABEL: . R inst/unitTests man tests In-Reply-To: <20110330235408.30F399C642@r-forge.r-project.org> References: <20110330235408.30F399C642@r-forge.r-project.org> Message-ID: Dear All, In this commit I have added patch of Nicola Pirastu + regression tests. All works fine, R CMD check GenABEL passed. I close bugs [#1184], [#1185], [#1259]. This is indeed a great patch, which kills three bugs at once! Many thanks, Nicola! We should keep in mind to update the topics on forum.genabel.org adding a message that the problem is fixed in 1.6-6 (now @RForge), which, when released on CRAN, will contain Nicola's patch. Nicola, you are the first one to contribute a patch to genabel-devel list, and I wonder if you would find it possible to draft a small "HOWTO make and contribute a GenABEL project patch"? best wishes, Yurii Details: Committing the patch of Nicola Pirastu http://lists.r-forge.r-project.org/pipermail/genabel-devel/2011-March/000182.html http://lists.r-forge.r-project.org/pipermail/genabel-devel/2011-March/000182.html > --- pkg/GenABEL/CHANGES.LOG ? ? 2011-03-30 16:33:24 UTC (rev 703) > +++ pkg/GenABEL/CHANGES.LOG ? ? 2011-03-30 23:54:07 UTC (rev 704) > @@ -1,5 +1,13 @@ > -*** v. 1.6-6 (2011.03.22) > +*** v. 1.6-6 (2011.03.31) > > +Applied the patch of Nicola Pirastu > +http://lists.r-forge.r-project.org/pipermail/genabel-devel/2011-March/000182.html > +to descriptives.trait. Added RUnit regression tests, updated > +documentation. Bugs fixed: [#1184], [#1185], [#1259] > + > Modified: pkg/GenABEL/R/descriptives.trait.R > =================================================================== > --- pkg/GenABEL/R/descriptives.trait.R ?2011-03-30 16:33:24 UTC (rev 703) > +++ pkg/GenABEL/R/descriptives.trait.R ?2011-03-30 23:54:07 UTC (rev 704) > @@ -1,60 +1,96 @@ > -"descriptives.trait" <- > -function (data,subset,file,by.var=NULL,digits=3) { > - ? ? ? if (missing(data)) stop("data argument must be provided") > - ? ? ? if (is(data,"gwaa.data")) data <- data at phdata > - ? ? ? if (!is(data,"data.frame")) stop("data argument must be of gwaa.data or data.frame-class") > - ? ? ? len <- dim(data)[1] > - ? ? ? ntra <- dim(data)[2] > - ? ? ? if (!missing(subset)) data <- data[subset,] > - ? ? ? if (!is.null(by.var)) { > - ? ? ? ? ? ? ? svar <- by.var > - ? ? ? ? ? ? ? if (length(levels(factor(svar)))!=2) stop("The by.var argument should contain a binary variable") > - ? ? ? ? ? ? ? out <- matrix(data=NA,nrow=ntra,ncol=9) > - ? ? ? ? ? ? ? lvls <- levels(factor(svar)) > - ? ? ? ? ? ? ? for (i in (1:ntra)) { > - ? ? ? ? ? ? ? ? ? ? ? ctrao <- data[,i] > - ? ? ? ? ? ? ? ? ? ? ? if (!is.numeric(ctrao) | all(ctrao==svar)) { > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ctra <- ctrao[which(svar == lvls[1])] > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,1] <- length(ctra) - sum(is.na(ctra)) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ctra <- ctrao[which(svar == lvls[2])] > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,4] <- length(ctra) - sum(is.na(ctra)) > - ? ? ? ? ? ? ? ? ? ? ? } else { > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ctra <- ctrao[which(svar == lvls[1])] > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,1] <- length(ctra) - sum(is.na(ctra)) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,2] <- mean(ctra,na.rm=TRUE) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,3] <- sd(ctra,na.rm=TRUE) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ctra <- ctrao[which(svar == lvls[2])] > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,4] <- length(ctra) - sum(is.na(ctra)) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,5] <- mean(ctra,na.rm=TRUE) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,6] <- sd(ctra,na.rm=TRUE) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,7] <- t.test(ctrao ~ svar)$p.value > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? out[i,8] <- kruskal.test(ctrao ~ svar)$p.value > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? clv <- length(unique(ctrao)) > - ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? if (clv>1 & clv<5) out[i,9] <- fisher.test(ctrao,svar)$p.value > - ? ? ? ? ? ? ? ? ? ? ? } > - ? ? ? ? ? ? ? } > - ? ? ? } else { > - ? ? ? ? ? ? ? out <- matrix(data=NA,nrow=ntra,ncol=3) > - ? ? ? ? ? ? ? for (i in (1:ntra)) { > - ? ? ? ? ? ? ? ? ? ? ? ctra <- data[,i] > - ? ? ? ? ? ? ? ? ? ? ? out[i,1] <- length(ctra) - sum(is.na(ctra)) > - ? ? ? ? ? ? ? ? ? ? ? if (!is.numeric(ctra)) next; > - ? ? ? ? ? ? ? ? ? ? ? out[i,2] <- mean(ctra,na.rm=TRUE) > - ? ? ? ? ? ? ? ? ? ? ? out[i,3] <- sd(ctra,na.rm=TRUE) > - ? ? ? ? ? ? ? } > - ? ? ? } > - ? ? ? out <- round(out,digits=digits) > - ? ? ? out <- data.frame(out) > - ? ? ? rownames(out) <- colnames(data) > - ? ? ? if (is.null(by.var)) > - ? ? ? ? ? ? ? colnames(out) <- c("No","Mean","SD") > - ? ? ? else > - ? ? ? ? ? ? ? colnames(out) <- c(paste("No(by.var=",lvls[1],")",sep=""),"Mean","SD",paste("No(by.var=",lvls[2],")",sep=""),"Mean","SD","Ptt","Pkw","Pexact") > - ? ? ? if (!missing(file)) { > - ? ? ? ? ? ? ? cat("\t",file=file,sep="") > - ? ? ? ? ? ? ? cat(colnames(out),file=file,sep="\t",append=TRUE) > - ? ? ? ? ? ? ? cat("\n",file=file,sep="",append=TRUE) > - ? ? ? ? ? ? ? write.table(out,file=file,sep="\t",append=T,col.names=FALSE) > - ? ? ? } > - ? ? ? out > +descriptives.trait <- function (data, subset, file, by.var = NULL, digits = 3) > +{ > + ? ?if (missing(data)) > + ? ? ? ?stop("data argument must be provided") > + ? ?if (is(data, "gwaa.data")) > + ? ? ? ?data <- data at phdata > + ? ?if (!is(data, "data.frame")) > + ? ? ? ?stop("data argument must be of gwaa.data or data.frame-class") > + ? ?len <- dim(data)[1] > + ? ?ntra <- dim(data)[2] > + ? ?if (!missing(subset)) > + ? ? ? ?data <- data[subset, ] > + ? ?if (!is.null(by.var)) { > + ? ? ? svar <- by.var > + ? ? ? ?if(class(by.var)=="formula"){ > + ? ? ? ? ? svar<-data[,as.character(by.var)[2]] > + ? ? ? ?} > + ? ? ? if(is.character(by.var) & length(by.var==1)){ > + ? ? ? ? ?svar<-data[,by.var] > + ? ? ? } > + ? ? ? ?if (length(levels(factor(svar))) != 2) > + ? ? ? ? ? ?stop("The by.var argument should contain a binary variable") > + ? ? ? ?out <- matrix(data = NA, nrow = ntra, ncol = 9) > + ? ? ? ?lvls <- levels(factor(svar)) > + ? ? ? ?for (i in (1:ntra)) { > + ? ? ? ? ? ?ctrao <- data[, i] > + ? ? ? ? ? ?if (!is.numeric(ctrao) | all(ctrao == svar,na.rm=T)) { > + ? ? ? ? ? ? ? ?ctra <- ctrao[svar == lvls[1]] > + ? ? ? ? ? ? ? ?out[i, 1] <- length(ctra) - sum(is.na(ctra)) > + ? ? ? ? ? ? ? ?ctra <- ctrao[svar == lvls[2]] > + ? ? ? ? ? ? ? ?out[i, 4] <- length(ctra) - sum(is.na(ctra)) > + ? ? ? ? ? ?} > + ? ? ? ? ? ?else { > + ? ? ? ? ? ? ? ?ctra <- ctrao[svar == lvls[1]] > + ? ? ? ? ? ? ? ?out[i, 1] <- length(ctra) - sum(is.na(ctra)) > + ? ? ? ? ? ? ? ?out[i, 2] <- mean(ctra, na.rm = TRUE) > + ? ? ? ? ? ? ? ?out[i, 3] <- sd(ctra, na.rm = TRUE) > + ? ? ? ? ? ? ? ?ctra <- ctrao[svar == lvls[2]] > + ? ? ? ? ? ? ? ?out[i, 4] <- length(ctra) - sum(is.na(ctra)) > + ? ? ? ? ? ? ? ?out[i, 5] <- mean(ctra, na.rm = TRUE) > + ? ? ? ? ? ? ? ?out[i, 6] <- sd(ctra, na.rm = TRUE) > + ? ? ? ? ? ? ? ?tmp<-try(t.test(ctrao ~ svar)$p.value) > + ? ? ? ? ? ? ? ?if(class(tmp)=="numeric"){ > + ? ? ? ? ? ? ? ? ? out[i, 7] <- tmp > + ? ? ? ? ? ? ? ?} > + ? ? ? ? ? ? ? ?tmp<-try(kruskal.test(ctrao ~ svar)$p.value) > + ? ? ? ? ? ? ? ?if(class(tmp)=="numeric"){ > + ? ? ? ? ? ? ? ? ? out[i, 8] <- tmp > + ? ? ? ? ? ? ? ?} > + ? ? ? ? ? ? ? ?clv <- length(unique(ctrao)) > + ? ? ? ? ? ? ? ?if (clv > 1 & clv < 5) > + ? ? ? ? ? ? ? ? ?out[i, 9] <- fisher.test(ctrao, svar)$p.value > + ? ? ? ? ? ?} > + ? ? ? ?} > + ? ?} > + ? ?else { > + ? ? ? ?out <- matrix(data = NA, nrow = ntra, ncol = 3) > + ? ? ? ?for (i in (1:ntra)) { > + ? ? ? ? ? ?ctra <- data[, i] > + ? ? ? ? ? ?out[i, 1] <- length(ctra) - sum(is.na(ctra)) > + ? ? ? ? ? ?if (!is.numeric(ctra)) > + ? ? ? ? ? ? ? ?next > + ? ? ? ? ? ?out[i, 2] <- mean(ctra, na.rm = TRUE) > + ? ? ? ? ? ?out[i, 3] <- sd(ctra, na.rm = TRUE) > + ? ? ? ?} > + ? ?} > + ? ?out <- round(out, digits = digits) > + ? ?out <- data.frame(out) > + ? ?rownames(out) <- colnames(data) > + ? ?if (is.null(by.var)) > + ? ? ? ?colnames(out) <- c("No", "Mean", "SD") > + ? ?else colnames(out) <- c(paste("No(by.var=", lvls[1], ")", > + ? ? ? ?sep = ""), "Mean", "SD", paste("No(by.var=", lvls[2], > + ? ? ? ?")", sep = ""), "Mean", "SD", "Ptt", "Pkw", "Pexact") > + ? ?if (!missing(file)) { > + ? ? ? ?cat("\t", file = file, sep = "") > + ? ? ? ?cat(colnames(out), file = file, sep = "\t", append = TRUE) > + ? ? ? ?cat("\n", file = file, sep = "", append = TRUE) > + ? ? ? ?write.table(out, file = file, sep = "\t", append = T, > + ? ? ? ? ? ?col.names = FALSE) > + ? ?} > + ? ?out > ?} > + Added regression tests: > Added: pkg/GenABEL/inst/unitTests/runit.descriptives.trait.R > =================================================================== > --- pkg/GenABEL/inst/unitTests/runit.descriptives.trait.R ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? (rev 0) > +++ pkg/GenABEL/inst/unitTests/runit.descriptives.trait.R ? ? ? 2011-03-30 23:54:07 UTC (rev 704) > @@ -0,0 +1,46 @@ > +### --- Test setup --- > +# > +# regression tests > +# > + > +if(FALSE) { > + ? ?## Not really needed, but can be handy when writing tests > + ? ?library(RUnit) > + ? ?library(GenABEL) > +} > + > +### do not run > +#stop("SKIP THIS TEST") > +### > + > +### ---- common functions and data ----- > + > +#source(paste("../inst/unitTests/shared_functions.R")) > +#source(paste(path,"/shared_functions.R",sep="")) > + > +### --- Test functions --- > + > +test.descriptives.trait <- function() > +{ > + ? ?data(ge03d2ex) > +# this works > + ? ?descriptives.trait(ge03d2ex,by.var=phdata(ge03d2ex)$sex) > +# bug [#1184] > +# and this does not! > + ? ?checkException(descriptives.trait(ge03d2ex,by.var=sex)) > + ? ?attach(phdata(ge03d2ex)) > + ? ?descriptives.trait(ge03d2ex,by.var=sex) > + ? ?detach(phdata(ge03d2ex)) > + ? ?descriptives.trait(ge03d2ex,by.var="sex") > + ? ?phdata(ge03d2ex)$sex[2] <- NA > +# bug [#1185] > +# and this does not! > + ? ?descriptives.trait(ge03d2ex,by.var=phdata(ge03d2ex)$sex) > + ? ?descriptives.trait(ge03d2ex,by.var="sex") > +# bug [#1259] > + ? ?convert.snp.illumina(infile="test_markers", outfile="test.raw") > + ? ?test = load.gwaa.data(pheno="test_phenos", geno="test.raw", force=T) > + ? ?attach(phdata(test)) > + ? ?descriptives.trait(data=test, by = bt) > + ? ?detach(phdata(test)) > +} > \ No newline at end of file > Updated documentation: > Modified: pkg/GenABEL/man/descriptives.trait.Rd > =================================================================== > --- pkg/GenABEL/man/descriptives.trait.Rd ? ? ? 2011-03-30 16:33:24 UTC (rev 703) > +++ pkg/GenABEL/man/descriptives.trait.Rd ? ? ? 2011-03-30 23:54:07 UTC (rev 704) > @@ -12,7 +12,9 @@ > ? \item{subset}{Subset of people to run analysis on. > ? ? ? ? ? ? ? ?If missing, all people from \code{data} are used for analysis.} > ? \item{file}{A string specifying the name of a file to write the tables to (default is no file otput).} > - ?\item{by.var}{a binary trait; statistics will be produced seprately for the groups and compared} > + ?\item{by.var}{a binary variable or a character scalar specifying the name > + ?of a binary trait in data; statistics will be produced separately for the > + ?groups and compared} > ? \item{digits}{number of digits to be printed} > ?} > ?%\details{ > @@ -31,5 +33,9 @@ > ? ? ? ?data(srdta) > ? ? ? ?descriptives.trait(srdta) > ? ? ? ?descriptives.trait(srdta,by.var=srdta at phdata$sex) > + ? ? ? descriptives.trait(srdta,by.var="sex") > + ? ? ? attach(phdata(srdta)) > + ? ? ? descriptives.trait(srdta,by.var=sex) > + ? ? ? detach(phdata(srdta)) > ?} > ?\keyword{distribution} > From yurii.aulchenko at gmail.com Thu Mar 31 02:27:40 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 31 Mar 2011 02:27:40 +0200 Subject: [GenABEL-dev] void error(char) function definition, where to find it? In-Reply-To: <1301224352.2781.16.camel@barabas> References: <1301224352.2781.16.camel@barabas> Message-ID: Just to keep the group updated: As we have discussed today with Lennart, this function is really meant for filevector, which can be compiled in two modes (as part of an R lib and as standalone, see code of frerror.*). So the logical solution for standalone ProbABEL is to replace this function with [ fprintf(errormessage); exit; ] Lennart will hopefully do the fix as time permits. Yurii On Sun, Mar 27, 2011 at 1:12 PM, L.C. Karssen wrote: > Hi all, > > I've been trying to fix ProbABEL bug #1339 [1] (problem that ProbABEL > doesn't compile because frerror.h is missing). This missing header file > (and the accompanying .cpp file) define a function > void error(const char * format, ... ); > > The frerror files have been deleted from the fvlib directory in svn in > r599, but grep-ing the complete genabel/pkg directory I still see that > several parts of the code call a function error(char), for example > Yurii's convert_snp_merlin.cpp (GenABEL) and Maksim's > supplementary_functions.cpp (VariABEL). Since I couldn't find a > definition of the function in the code of the GenABEL suite, I assume it > is defined elsewhere. > > I've tried to find out where this function is defined, but I failed. > Who can point me to the (header) file where the error() function is > defined? > > > Thanks, > > Lennart. > > > [1] > https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1339&group_id=505&atid=2058 > > -- > ----------------------------------------------------------------- > L.C. Karssen > Utrecht > The Netherlands > > lennart at karssen.org > http://blog.karssen.org > > Stuur me aub geen Word of Powerpoint bestanden! > Zie http://www.gnu.org/philosophy/no-word-attachments.nl.html > ------------------------------------------------------------------ > > _______________________________________________ > genabel-devel mailing list > genabel-devel at lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel > > From pirastu at burlo.trieste.it Thu Mar 31 12:20:28 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Thu, 31 Mar 2011 12:20:28 +0200 (CEST) Subject: [GenABEL-dev] descriptives.trait Patch In-Reply-To: References: <49674.193.62.206.85.1301054540.squirrel@webmail-rupar.regione.fvg.it> Message-ID: <50477.212.189.217.13.1301566828.squirrel@webmail-rupar.regione.fvg.it> Hi Yurii, I was looking at the documentation changes you made, and I just wanted to point out that another version of the command is: descriptives.trait(srdta, by.var=~sex) whithout "". Have no idea if this is usefull, but anyways just to let you know it was in there. Best Nicola > Dear Nicola, > > Many thanks for submitting this patch! These little problems were > pestering 'descriptive.traits' from the very moment the procedure was > invented -- so it is great news indeed and a great relief to have > these problems fixed! > > I see no problem with applying this patch to the main release (and > within 5 days there were no concerns from others). So, I will apply > this patch + add RUnit regression tests for these three bugs -- and > when the commit (or any problems) will be there you will hear about > this on the mailing list. > > best wishes, > Yurii > > On Fri, Mar 25, 2011 at 1:02 PM, wrote: >> Dear all, >> >> attached you can find a patch to fix the bugs described underneath. To >> install from the root directory of /pkg: patch -p0 < >> patch_descriptives.trait. >> >> Best >> >> Nicola >> >> Patch for bugs #1184 #1185 #1259 >> >> Bug [#1184] descriptives.trait can not see the variable in by.var >> argument >> unless full path specified >> >> Description: >> https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1184&group_id=505&atid=2058 >> >> Changes: now by.var can be specified aslo as :"var_name" or ~var_name >> >> Test: >> >>> library(GenABEL) >>> data(ge03d2ex) >>> descriptives.trait(ge03d2ex,by.var="sex") >> ? ? ? No(by.var=0) ? ?Mean ? ? SD No(by.var=1) ? ?Mean ? ? SD ? Ptt ? >> Pkw >> id ? ? ? ? ? ? ? 64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? >> ?NA >> sex ? ? ? ? ? ? ?64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? >> ?NA >> age ? ? ? ? ? ? ?64 ?46.942 12.479 ? ? ? ? ? 72 ?50.959 13.107 0.070 >> 0.081 >> dm2 ? ? ? ? ? ? ?64 ? 0.547 ?0.502 ? ? ? ? ? 72 ? 0.708 ?0.458 0.053 >> 0.052 >> height ? ? ? ? ? 64 162.680 ?6.819 ? ? ? ? ? 71 175.534 ?7.943 0.000 >> 0.000 >> weight ? ? ? ? ? 64 ?78.605 26.908 ? ? ? ? ? 71 ?95.322 21.441 0.000 >> 0.000 >> diet ? ? ? ? ? ? 64 ? 0.109 ?0.315 ? ? ? ? ? 72 ? 0.014 ?0.118 0.025 >> 0.019 >> bmi ? ? ? ? ? ? ?64 ?29.604 ?9.506 ? ? ? ? ? 71 ?30.930 ?6.547 0.352 >> 0.040 >> ? ? ? Pexact >> id ? ? ? ? NA >> sex ? ? ? ?NA >> age ? ? ? ?NA >> dm2 ? ? 0.074 >> height ? ? NA >> weight ? ? NA >> diet ? ?0.026 >> bmi ? ? ? ?NA >>> descriptives.trait(ge03d2ex,by.var=~sex) >> ? ? ? No(by.var=0) ? ?Mean ? ? SD No(by.var=1) ? ?Mean ? ? SD ? Ptt ? >> Pkw >> id ? ? ? ? ? ? ? 64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? >> ?NA >> sex ? ? ? ? ? ? ?64 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? >> ?NA >> age ? ? ? ? ? ? ?64 ?46.942 12.479 ? ? ? ? ? 72 ?50.959 13.107 0.070 >> 0.081 >> dm2 ? ? ? ? ? ? ?64 ? 0.547 ?0.502 ? ? ? ? ? 72 ? 0.708 ?0.458 0.053 >> 0.052 >> height ? ? ? ? ? 64 162.680 ?6.819 ? ? ? ? ? 71 175.534 ?7.943 0.000 >> 0.000 >> weight ? ? ? ? ? 64 ?78.605 26.908 ? ? ? ? ? 71 ?95.322 21.441 0.000 >> 0.000 >> diet ? ? ? ? ? ? 64 ? 0.109 ?0.315 ? ? ? ? ? 72 ? 0.014 ?0.118 0.025 >> 0.019 >> bmi ? ? ? ? ? ? ?64 ?29.604 ?9.506 ? ? ? ? ? 71 ?30.930 ?6.547 0.352 >> 0.040 >> ? ? ? Pexact >> id ? ? ? ? NA >> sex ? ? ? ?NA >> age ? ? ? ?NA >> dm2 ? ? 0.074 >> height ? ? NA >> weight ? ? NA >> diet ? ?0.026 >> bmi ? ? ? ?NA >> >> Bug [#1185] descriptives.trait with missings in by.var argument >> >> Description: >> https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1185&group_id=505&atid=2058 >> >> Fixed: now by.var argument acceptes missing >> >> Test: >> >>> phdata(ge03d2ex)$sex[2] <- NA >>> descriptives.trait(ge03d2ex,by.var=phdata(ge03d2ex)$sex) >> ? ? ? No(by.var=0) ? ?Mean ? ? SD No(by.var=1) ? ?Mean ? ? SD ? Ptt ? >> Pkw >> id ? ? ? ? ? ? ? 63 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? >> ?NA >> sex ? ? ? ? ? ? ?63 ? ? ?NA ? ? NA ? ? ? ? ? 72 ? ? ?NA ? ? NA ? ?NA ? >> ?NA >> age ? ? ? ? ? ? ?63 ?47.003 12.570 ? ? ? ? ? 72 ?50.959 13.107 0.076 >> 0.088 >> dm2 ? ? ? ? ? ? ?63 ? 0.540 ?0.502 ? ? ? ? ? 72 ? 0.708 ?0.458 0.045 >> 0.044 >> height ? ? ? ? ? 63 162.573 ?6.819 ? ? ? ? ? 71 175.534 ?7.943 0.000 >> 0.000 >> weight ? ? ? ? ? 63 ?77.651 26.009 ? ? ? ? ? 71 ?95.322 21.441 0.000 >> 0.000 >> diet ? ? ? ? ? ? 63 ? 0.111 ?0.317 ? ? ? ? ? 72 ? 0.014 ?0.118 0.024 >> 0.017 >> bmi ? ? ? ? ? ? ?63 ?29.306 ?9.277 ? ? ? ? ? 71 ?30.930 ?6.547 0.250 >> 0.025 >> ? ? ? Pexact >> id ? ? ? ? NA >> sex ? ? ? ?NA >> age ? ? ? ?NA >> dm2 ? ? 0.051 >> height ? ? NA >> weight ? ? NA >> diet ? ?0.025 >> bmi ? ? ? ?NA >> >> Bug [#1259] descriptives.trait with by.var argument does not work >> (contributed by Richard Ahn) >> >> Description: >> https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1259&group_id=505&atid=2058 >> >> Comments: The problem was that for one of the traits one of the >> categories >> had all missing data, this made descriptives.trait crush. >> >> Fixed: descriptives.trait does not crush anymore. >> >> Test: >> >> nicola at nicola-VirtualBox:~/Scrivania$ R -f test.R >> WARNING: ignoring environment value of R_HOME >> >> R version 2.11.1 (2010-05-31) >> Copyright (C) 2010 The R Foundation for Statistical Computing >> ISBN 3-900051-07-0 >> >> R ? un software libero ed ? rilasciato SENZA ALCUNA GARANZIA. >> Siamo ben lieti se potrai redistribuirlo, ma sotto certe condizioni. >> Scrivi 'license()' o 'licence()' per dettagli su come distribuirlo. >> >> R ? un progetto di collaborazione con molti contributi esterni. >> Scrivi 'contributors()' per maggiori informazioni e 'citation()' >> per sapere come citare R o i pacchetti di R nelle pubblicazioni. >> >> Scrivi 'demo()' per una dimostrazione, 'help()' per la guida in linea, o >> 'help.start()' per l'help navigabile con browser HTML. >> Scrivi 'q()' per uscire da R. >> >>> library(GenABEL) >> Carico il pacchetto richiesto: MASS >> GenABEL v. 1.6-6 (February 24, 2011) loaded >> >> Installed GenABEL version (1.6-6) is not the same as stable >> version available from CRAN (1.6-5). Unless used intentionally, >> consider updating to the latest CRAN version. For that, use >> 'install.packages("GenABEL")', or ask your system administrator >> to update the package. >> >>> convert.snp.illumina(infile="test_markers", outfile="test.raw") >> Reading genotypes from file 'test_markers' ... >> Writing to file 'test.raw' ... >> ... done. >>> test = load.gwaa.data(pheno="test_phenos", geno="test.raw", force=T) >> ids loaded... >> marker names loaded... >> chromosome data loaded... >> map data loaded... >> allele coding data loaded... >> strand data loaded... >> genotype data loaded... >> snp.data object created... >> assignment of gwaa.data object FORCED; X-errors were not checked! >>> attach(phdata(test)) >>> descriptives.trait(data=test, by = bt) >> Error in t.test.formula(ctrao ~ svar) : >> ?grouping factor must have exactly 2 levels >> Error in kruskal.test.default(c(6L, 32L, 35L, 33L, 52L, 25L), c(1L, 1L, >> ?: >> ?all observations are in the same group >> ? ?No(by.var=0) Mean ? ?SD No(by.var=1) ? Mean ? ? SD ? Ptt ? Pkw Pexact >> id ? ? ? ? ? ? 5 ? NA ? ?NA ? ? ? ? ? ?6 ? ? NA ? ? NA ? ?NA ? ?NA ? ? >> NA >> sex ? ? ? ? ? ?5 ?0.2 0.447 ? ? ? ? ? ?6 ?0.333 ?0.516 0.657 0.637 ? ? >> ?1 >> age ? ? ? ? ? ?0 ?NaN ? ?NA ? ? ? ? ? ?6 30.500 14.977 ? ?NA ? ?NA ? ? >> NA >> bt ? ? ? ? ? ? 5 ? NA ? ?NA ? ? ? ? ? ?6 ? ? NA ? ? NA ? ?NA ? ?NA ? ? >> NA >>> >>> >> >> >> >> >> >> >> >> _______________________________________________ >> genabel-devel mailing list >> genabel-devel at lists.r-forge.r-project.org >> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel >> > From yurii.aulchenko at gmail.com Thu Mar 31 13:56:08 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 31 Mar 2011 13:56:08 +0200 Subject: [GenABEL-dev] descriptives.trait Patch In-Reply-To: <50477.212.189.217.13.1301566828.squirrel@webmail-rupar.regione.fvg.it> References: <49674.193.62.206.85.1301054540.squirrel@webmail-rupar.regione.fvg.it> <50477.212.189.217.13.1301566828.squirrel@webmail-rupar.regione.fvg.it> Message-ID: > I was looking at the documentation changes you made, and I just wanted to > point out that another version of the command is: > > descriptives.trait(srdta, by.var=~sex) > > whithout "". I have noticed that type of passing an argument in your test-code, but thought of dropping this as I have no idea what this means. Looking for help("~") returned explanation for "~" as a part of formula. Do you know what this actually means and how it works? Yurii From pirastu at burlo.trieste.it Thu Mar 31 17:15:20 2011 From: pirastu at burlo.trieste.it (pirastu at burlo.trieste.it) Date: Thu, 31 Mar 2011 17:15:20 +0200 (CEST) Subject: [GenABEL-dev] [Fwd: Re: descriptives.trait Patch] Message-ID: <36652.140.105.110.180.1301584520.squirrel@webmail-rupar.regione.fvg.it> Actually yes, its a trick I've used to eliminate the "". If you use ~ R will interpeter as a formula and will not require quotes. You can try to type something like a<-x~y, a will be of class formula. you can actually make lists with more formulas. a<- x~y b<-f~d c(a,b) They will be taken by lm and other things. That said this will work also with ~x for example. If you do as.character(a) [1] "~" "x" "y" basically I used this to avoid having the error " sex is not an object" or something like that. Also from a "formal" point of view it makes sense, in fact when you make a boxplot for example you would type boxplot(trait~sex) to get a boxplot for each category. I could infact change descriptives.trait to work in the same way, something like descriptives.trait(srdta~sex) which might be a little more intuitive, but this solution was quicker at least to implement. Hope to have answered your questions. Best Nicola >> I was looking at the documentation changes you made, and I just wanted >> to >> point out that another version of the command is: >> >> descriptives.trait(srdta, by.var=~sex) >> >> whithout "". > > I have noticed that type of passing an argument in your test-code, but > thought of dropping this as I have no idea what this means. Looking > for help("~") returned explanation for "~" as a part of formula. > > Do you know what this actually means and how it works? > > Yurii > From yurii.aulchenko at gmail.com Thu Mar 31 21:08:55 2011 From: yurii.aulchenko at gmail.com (Yurii Aulchenko) Date: Thu, 31 Mar 2011 21:08:55 +0200 Subject: [GenABEL-dev] [Fwd: Re: descriptives.trait Patch] In-Reply-To: <36652.140.105.110.180.1301584520.squirrel@webmail-rupar.regione.fvg.it> References: <36652.140.105.110.180.1301584520.squirrel@webmail-rupar.regione.fvg.it> Message-ID: Nicola, Thanks for explanation! > I could infact change descriptives.trait to work in the same way, > something like > > descriptives.trait(srdta~sex) > > which might be a little more intuitive, but this solution was quicker at > least to implement. > I think this is very good suggestion, and this indeed looks like a more 'conventional way I can not see any reason why not to do this (still, for backward compatibility, it is important to keep "by.var" argument functional as well) best wishes, Yurii