in reply to Re: phylogenetic tree construction using perl
in thread phylogenetic tree construction using perl

Hi greengaroo, The algorithm is this:- Pictorial representation:- http://picpaste.com/triplets-IQMFT1QY.jpg 
Triplets :: S=('b,c|a', 'a,c|d', 'd,e|b'),
Species :: L={a,b,c,d,e}
TreeConstruct(S):
1.] Let L be the set of species in S. Build G(L) the auxillary graph.
2.] Let C1,C2....Cq be the set of connected components in G(L).
3.] If q>1,then
   - For i=1,2.....q, let S(i) be the set of triplets in S labeled by 
+the set of leaves in C(i).
   - Let T(i) = TreeConstruct(S(i))
   - Let T be a tree formed by connecting all T(i) with the same paren
+t node. Return T.
4.]If q=1 & C1 contains exactly one leaf,return the leaf ,else return 
+fail.

[download]
I have updated the code and now it takes input connections in form of triplets and prints the connected components of the graph. 
use strict;  
use warnings;   
use Graph;  
@ARGV =  ('b,c|a', 'a,c|d', 'd,e|b') unless @ARGV;   
  
my %HoA;  
foreach ( @ARGV ) {  
    m/^([a-z])[,]([a-z])[|]([a-z])$/ ;  
    push @{$HoA{$1}}, $2;  
}  
print "\n===========\@HoA=====\n";  
print "from->to\n";  
while (my ($key, $values) = each %HoA) {  
    print $key, "=>   [", join(',', @$values), "]\n";  
}  


my $g = Graph->new( undirected => 1 );

for my $src ( keys %HoA ) {
    for my $tgt ( @{ $HoA{$src} } ) {
        $g->add_edge($src, $tgt);
    }
}

my @subgraphs = $g->connected_components;
my @allgraphs;

for my $subgraph ( @subgraphs ) {
    push @allgraphs, {};
    for my $node ( @$subgraph ) {
        if ( exists $HoA{ $node } ) {
            $allgraphs[-1]{$node} = [ @{ $HoA{$node} } ];
        }
    }
}
print "----connected components------------";
use YAML; print Dump \@allgraphs;
-------------OUTPUT----------------
===========@HoA=====
from->to
a=>   [c]
b=>   [c]
d=>   [e]
----connected components---------------
- a:
    - c
  b:
    - c
- d:
    - e

[download]
Hope this helps you get an idea

Replies are listed 'Best First'.
Re^3: phylogenetic tree construction using perl
by jgamble (Pilgrim) on Sep 27, 2012 at 18:45 UTC

    I'm still reading through the Triplet Methods paper, so I'm certain my understanding is not only limited but wrong. Nonetheless, section 7.4.1 ("Reconstruct a network by a sorting network") caught my eye for the simple reason that Algorithm-Networksort exists on CPAN and I am its author.

    So if the module can be of use to you, great. If there's a feature that you need from it that's doable, I'd be more than happy to add it to the module. Let me know.

    By the way, your link to your jpeg on picpaste doesn't display anything.

[reply]

      Thanks for pointing it out jgamble. Actually the picpaste had figure 7.4 from the paper (which depicts the algorithm for a better understanding). And the algorithm is Alfred Aho's famous algorithm, which has been adopted by biologists to construct a tree. The problem is that it deals with connected components, subgraphs,recursion etc all at once. So Im having trouble proceeding, though im still trying to solve it moving in bits and pieces that is.

      The algorithm TreeConstruct described in section 7.2.3 is my main concern.
[reply]
      Hi jgamble, Did you get the algorithm. Please help me on it. http://picpaste.com/Z1SVFTT6.png

      This is what a triplet looks like. The link below shows three triplets (a,b|c) (a,c|d) (d,e|b) and their consensus supertree. http://picpaste.com/Nu0ON9uo.jpg

[reply]

        I'm only able to look at the paper in my spare time, and I had no spare time for the past few days. Also, quite frankly, I didn't know if your problem had been solved already or not.

        Did the bioperl suggestions lead anywhere for you?

        I'll resume reading the paper tomorrow. If you have any further reading material that might be helpful, please let me know. I can be reached more directly at jgamble@cpan.org.

[reply]

In Section Seekers of Perl Wisdom