[Phylobase-devl] nTips method defined where?

[Ben Bolker]

   Without thinking about it too carefully (on my way out
my office door), I can imagine that reticulation is one
of the things we'd like to allow wherever possible.  I have
no problem with putting in a more robust test that
handles this case, but I do predict that you will run
into other funny cases that might be harder to handle.
(As I said before, though, I consider this all a good
part of shaking down the structure of phylobase.)

   I will try to commit this tomorrow ...

   Ben

Aaron Mackey wrote:
> Here's my edge matrix:
> 
>       ancestor descendant
>  [1,]        6          7
>  [2,]        6          1
>  [3,]        7          8
>  [4,]        7          9
>  [5,]        8          2
>  [6,]        8          3
>  [7,]        9          4
>  [8,]        9          5
>  [9,]        8          4
> 
> note the last edge implies that node 4 has two ancestors (i.e. this is a
> reticulated network, not a strict tree).
> 
>> sum(tabulate(F[,1]) == 0)
> [1] 5
>> sum(!(F[,2] %in% F[,1]))
> [1] 6
> 
> A quick fix without changing the idiom would be:
>> sum(!(unique(F[,2]) %in% F[,1]))
> [1] 5
> 
> 
> On Wed, May 21, 2008 at 3:20 PM, Ben Bolker <bolker at zoology.ufl.edu> wrote:
> 
>>  It's in methods-phylo4.R.  There have been some
>> issues in the past with conflicts between an "nTips"
>> function in ape and the one in phylo4, but I don't think
>> that's what's going on here.
>>
>> Our current definition for phylo4 objects is
>>
>> setMethod("nTips", "phylo4", function(x, ...) {
>>    E <- edges(x)
>>    res <- sum(!E[, 2] %in% E[, 1])
>>    return(res)
>> })
>>
>>  which, in words, is "number of nodes in column 2 that do
>> not appear in column 1".  To be honest, I don't remember
>> how we came up with this.  It would seem like "number
>> of nodes of (out)degree 0" would be a better definition --
>> i.e., something using the same tabulate() idiom
>> that you've seen already -- although
>> I'm having a hard time at the moment thinking of the
>> circumstances under which they would differ.
>>
>>  What does your edge matrix look like?  Can you see
>> why it would be different?  Does it violate any of
>> the ape rules quoted earlier?
>>
>>  Ben
>>
>>
>>
>>
>>  It's really nice to have someone else hacking on the code,
>> to flush out stuff like this!
>>
>> Aaron Mackey wrote:
>>
>>> When reading a nexml-formatted tree with reticulated nodes (i.e. a
>>> network),
>>> I'm getting an error while constructing the phylo4 tree that the number of
>>> tips doesn't match the number of tip labels.  This is stemming from
>>> check_phylo4, which checks that nTips(object) is equal to
>>> length(tip.label)
>>> -- but I can't find the code for nTips (here is a snippet of my debugging
>>> session within check_phylo4; there are only 5 tips, but nTips reports 6):
>>>
>>> Browse[1]> nTips(object)
>>> [1] 6
>>> Browse[1]> object at tip.label
>>> [1] "Homo sapiens" "t3"           "t2"           "t5"
>>> "t4"
>>> Browse[1]> getMethod("nTips", where=object)
>>> Method Definition:
>>>
>>> function (x, ...)
>>> {
>>>    .local <- function (x)
>>>    {
>>>        if (class(x) == "phylo") {
>>>            Ntip(x)
>>>        }
>>>        else stop(paste("no 'nTips' method available for",
>>> deparse(substitute(x)),
>>>            "(class", class(x), ")"))
>>>    }
>>>    .local(x, ...)
>>> }
>>> <environment: 0x861e04c>
>>>
>>> Signatures:
>>>        x
>>> target  "ANY"
>>> defined "ANY"
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> _______________________________________________
>>> Phylobase-devl mailing list
>>> Phylobase-devl at lists.r-forge.r-project.org
>>>
>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/phylobase-devl
>>>
>>
>>
> 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 252 bytes
Desc: OpenPGP digital signature
Url : http://lists.r-forge.r-project.org/pipermail/phylobase-devl/attachments/20080521/57f91934/attachment-0001.pgp