[GenABEL-dev] Introducing OmicABELnoMM

[L.C. Karssen]

Dear list,

I've made some progress with compiling OmicABELnoMM. I've just uploaded
a working configure.ac.

I installed the following packages on my Ubuntu 13.10 system:
libopenblas-base
libopenblas-dev
liblapacke{,-dev}
liblapack-dev
liblapack3

If I then run
./configure LDFLAGS="-L/usr/lib/openblas-base"
make

Compilation works (lots of warnings though, due to -Wall) and a binary
is created. When running the example from the howtocompile.txt file I
get a bunch of the following warnings (although the program seems to run
fine with the example data):

OpenBLAS Warning : Detect OpenMP Loop and this application may hang.
Please rebuild the library with USE_OPENMP=1 option.


So, as I guessed, Ubuntu's libopenblas is not compiled with openMP support.
I've tried to created my own packages based on the instructions in
/usr/share/doc/libopenblas-base/README.Debian and adding USE_OPENMP=1 to
the make flags in the debian/rules file. That works, and after
recompiling the OpenBLAS warnings disappear.

On jenkins the build is still failing, because there is no lapacke
package for that machine...


Lennart.

On 18-12-13 11:13, L.C. Karssen wrote:
> Dear list,
> 
> Here's an update on my first steps with OmicABELnoMM after uploading it
> to SVN:
> 
> 1) I spent some time getting several files into better compliance with
> our coding style guide.
> 
> 2) I started implementing autoconf/automake. This is hampered by 3).
> 
> 3) I tried using the openblas package that comes with Debian/Ubuntu, but
> that doesn't seem to work. After digging a lot, I now think this is
> because the Debian package is not compiled with the USE_OPENMP=1 flag
> (which is needed according to Alvaro's instructions).
> I'm going to investigate two ways to solve this:
> - rebuild the Debian package with openMP support. If that helps I have a
> stronger case to try and change the Debian package
> - Include a copy of openblas 2.8.2 into the OmicABELnoMM package. I've
> already verified that that works (basically following Alvaro's
> instructions), but from a design point of view I don't like really it.
> It would clutter SVN with components we're not interested in
> maintaining. An alternative may be to have a
> 'download-and-compile-openblas' script.
> Let me know if you have any suggestions or comments.
> 
> 4) I've added OmicABELnoMM to Jenkins. It doesn't look good yet, given
> that all the builds so far failed, but that will only improve in the
> future :-)
> 
> 
> Best,
> 
> Lennart.
> 
> 
> On 16-12-13 15:05, L.C. Karssen wrote:
>> Dear list,
>>
>> As you may have noticed, I have committed a first version of
>> OmicABELnoMM to SVN on R-forge this morning.
>> OmicABELnoMM is the no-mixed-models version of OmicABEL, allowing rapid
>> genome-wide association analysis (linear regression), especially for
>> multiple traits ("omics" data).
>> OmicABELnoMM was written from scratch by Alvaro Jesus Frank so that the
>> algorithm is well tailored to the different mathematics. Alvaro works in
>> the same team at RWTH Aachen as Diego Fabregat Traver of OmicABEL fame,
>> under the supervision of Paolo Bientinesi, Yurii Aulchenko and,
>> recently, myself.
>>
>> Please note that OmicABELnoMM is still under development (thanks Yakov
>> and Sodbo for your contributions!). Most notably, work is still being
>> done on the exact output file format. At present the output is stored in
>> a filevector-format file. This will probably remain the same, but the
>> order in which the information (betas, standard errors) is stored may
>> change.
>>
>> These are the development plans we have for OmicABELnoMM in the near future:
>> - Implement autotools configuration
>> - Improve the output file format
>> - Add documentation
>>
>>
>> Best regards,
>>
>> Lennart.
>>
>>
>>
>>
>> _______________________________________________
>> 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
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20131219/25d6b195/attachment.sig>