Re^2: Problems with ensembl perl script

[bioinfo]

in reply to Re: Problems with ensembl perl script
in thread Problems with ensembl perl script

Thank you both for the help and for your time. I'm stuck with this problem and do not know how to fix it. I made this change that you propose tfrayner but still gives an error:

-------------------- EXCEPTION --------------------
MSG: [Bio::EnsEMBL::Compara::Member=HASH(0x9adfe40)] must be a Bio::En
+sEMBL::Compara::MethodLinkSpeciesSet object

STACK Bio::EnsEMBL::Compara::DBSQL::HomologyAdaptor::fetch_all_by_Meth
+odLinkSpeciesSet_orthology_type_subtype /home/david/src/ensembl-compa
+ra/modules/Bio/EnsEMBL/Compara/DBSQL/HomologyAdaptor.pm:430
STACK toplevel ortologos.pl:35
---------------------------------------------------

I started with perl and Ensembl recently and I do not quite understand yet. The documentation is here:

http://www.ensembl.org/info/docs/api/Pdoc/ensembl-compara/modules/Bio/EnsEMBL/Compara/DBSQL/HomologyAdaptor.html#POD7

So there should show only the orthologous of Mammalia which are one-to-one.This another code works and returns all species homologues, but I only want mammalian orthologues.

use strict;
    use warnings;
    use Bio::EnsEMBL::Registry;
    ## Load the registry automatically
    my $reg = "Bio::EnsEMBL::Registry";
    $reg->load_registry_from_url('mysql://[e-mail].
+org');
    ## Get the human gene adaptor
    my $human_gene_adaptor =
        $reg->get_adaptor("Homo sapiens", "core", "Gene");
    ## Get the compara member adaptor
    my $member_adaptor =
        $reg->get_adaptor("Compara", "compara", "Member");
    ## Get the compara homology adaptor
    my $homology_adaptor =
        $reg->get_adaptor("Compara", "compara", "Homology");



    my @BreastCANgenes = ('ABCA3','ABCB10','ABCB8','ACADM');

    open ("archivo", ">homologos.txt");
    my $gen;

    foreach $gen (@BreastCANgenes) {


    ## Get all existing gene object
    my $ctdp1_genes = $human_gene_adaptor->fetch_all_by_external_name(
+$gen);
    ## For each of these genes...
    foreach my $ctdp1_gene (@$ctdp1_genes) {
      ## Get the compara member
      my $member = $member_adaptor->fetch_by_source_stable_id(
          "ENSEMBLGENE", $ctdp1_gene->stable_id);
      ## Get all the homologues
      my $all_homologies = $homology_adaptor->fetch_all_by_Member($mem
+ber);
      ## For each homology
      foreach my $this_homology (@$all_homologies) {
        ## print the description (type of homology) and the
        ## subtype (taxonomy level of the event: duplic. or speciation
+)
        print $this_homology->description, " [", $this_homology->subty
+pe, "]\n";
        print archivo $this_homology->description, " [", $this_homolog
+y->subtype, "]\n";
        ## print the members in this homology
        my $members = $this_homology->get_all_Members();
        foreach my $this_member (@$members) {
          print archivo $this_member->source_name, " ",
        $this_member->stable_id, " (",
        $this_member->genome_db->name, ")\n"


        }
        print "\n";
      }
    }
    }
    close ("archivo");

[tfrayner]

Hmm; I think I see what you're trying to do here, and I can see why you'd be frustrated. I've checked out the latest ensembl-compara code from CVS and it looks as though the (poorly named) fetch_all_by_MethodLinkSpeciesSet_orthology_type_subtype method has been removed, and it doesn't seem to have been replaced by anything with equivalent functionality. Maybe I'm wrong, and I guess the Ensembl devs would be the best people to ask. Maybe they've had a DB schema change that meant the old SQL query used to filter on subtype is no longer valid? Anyway, it may be that the best you can do is filter the homologies yourself:

        ## For each homology
        foreach my $this_homology (@$all_homologies) {

            next unless $this_homology->subtype eq 'Mammalia';

... and then the rest of your loop from the second script. That works on my machine and is reasonably fast, at least for this small gene set. I have no idea whether it will scale to meet your final requirements, though.

Cheers
Tim