Fixed find_genes subroutine for new versions of the sequence files

svn path=/; revision=28

diff --git a/interactome_scripts/find_species.pl b/interactome_scripts/find_species.pl
index eaf444dc47f903abe6a36c4d19aa46725114d237..77183d383560e6e540149a9374047ef7284f9ed0 100755 (executable)
--- a/interactome_scripts/find_species.pl
+++ b/interactome_scripts/find_species.pl
@@ -94,75 +94,130 @@ sub find_species {
 
 sub find_gene {
        my $gene_header = $_[0];
+       $gene_header =~ s/^>//; # strip off the header line identifier, if it isn't already
        my $species = $_[1];
        my $gene;
-       if ($species eq "Ath") {
+       if ($species eq "Arabidopsis_lyrata") {
+               my ($name,$gene_id,$scaffold,$isomer) = split(/\|/, $gene_header);
+               $gene = $isomer;
+       }elsif ($species eq "Arabidopsis_thaliana") {
+               my ($name,$gene_id,$chrom,$isomer) = split(/\|/, $gene_header);
+               $gene = $isomer;
+       }elsif ($species eq "Batrachochytrium_distachyon") {
+               my ($gene_id,$transcript,$info) = split(/\s\|\s/, $gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Brachypodium_distachyon") {
                my ($name,$gene_id,$chrom,$isomer) = split(/\|/, $gene_header);
                $gene = $isomer;
-       }elsif ($species eq "Brachy") {
-               #uncomment for newer version of fasta file
-               #my ($gene_id,$chrom,$isomer,$name) = split(/\|/, $gene_header);
-               #$gene = $isomer;
-               #uncomment for old version of fasta file
-               $gene = $gene_header;
-       }elsif ($species eq "C_elegans") {
-               my ($gene_id,$temp) = split(/\|/, $gene_header);
+       }elsif ($species eq "Caenorhabditis_elegans") {
+               my ($locus_id,$unknown,$gene_id,$temp) = split(/\s+/, $gene_header);
                $gene = $gene_id; #???
-       }elsif ($species eq "Chlamy") {
-               my ($name,$locus_id,$scaff_id,$temp) = split(/\|/,$gene_header);
-               $gene = $locus_id; #???
-       }elsif ($species eq "Danio") {
+       }elsif ($species eq "Carica_papaya") {
+               my $supercontig = $gene_header;
+               $gene = $supercontig;
+       }elsif ($species eq "Chlamydomonas_reinhardtii") {
+               my ($name,$locus_id,$chrom,$transcript) = split(/\|/,$gene_header);
+               $gene = $transcript; #???
+       }elsif ($species eq "Cucumis_sativus") {
+               my ($name,$gene_id,$scaffold,$unknown) = split(/\|/,$gene_header);
+               print "$name\t$gene_id\t$scaffold\t$unknown\n";
+               $gene = $gene_id;
+       }elsif ($species eq "Danio_rerio") {
+               my ($gene_id,$transcript) = split(/\|/,$gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Drosophila_melanogaster") {
                my ($gene_id,$temp) = split(/\s/,$gene_header);
                $gene = $gene_id;
-       }elsif ($species eq "E_coli") {
+       }elsif ($species eq "Ectocarpus_siliculosus") {
+               my ($temp,$transcript,$source,$gene_id,$func) = split(/\|/, $gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Escherichia_coli") {
                $gene = $gene_header; #???
-       }elsif ($species eq "Fragaria") {
-               my ($gene_id, $mrna_id, $method, $length) = split(/\|/, $gene_header);
+       }elsif ($species eq "Fragaria_vesca") {
+               my ($gene_id, $mrna_id, $method, $length) = split(/\s\|\s/, $gene_header);
                $gene = $gene_id;
-       }elsif ($species eq "Glycine") {
-               my ($name,$locus_id,$scaff_id,$isomer) = split(/\|/,$gene_header);
+       }elsif ($species eq "Glycine_max") {
+               my ($name,$locus_id,$isomer,$chrom) = split(/\|/,$gene_header);
                $gene = $isomer;
-       }elsif ($species eq "Human") {
-               my ($gene_id,$temp) = split(/\s/,$gene_header);
+       }elsif ($species eq "Homo_sapiens") {
+               my ($protein,$type,$chrom,$gene_id,$transcript) = split(/\s/,$gene_header);
+               $gene_id =~ s/Gene\://;
                $gene = $gene_id;
-       }elsif ($species eq "Maize") {
-               $gene = $gene_header;
-       }elsif ($species eq "Mouse") {
-               my ($gene_id,$temp) = split(/\s/,$gene_header);
+       }elsif ($species eq "Laccaria_bicolor") {
+               my ($source,$spec,$gene_id,$scaffold) = split(/\|/, $gene_header);
                $gene = $gene_id;
-       }elsif ($species eq "Neurospora") {
-               my ($gene_id,$temp) = split(/\s/,$gene_header);
+       }elsif ($species eq "Magnaporthe_grissa") {
+               my ($transcript,$gene_id,$temp) = split(/\s\|\s/, $gene_header);
+               $gene = $transcript;
+       }elsif ($species eq "Manihot_esculenta") {
+               my ($name,$scaffold,$gene_id,$temp) = split(/\|/, $gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Medicago_truncatula") {
+               my ($name,$gene_id,$chrom,$temp) = split(/\|/, $gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Mimulus_guttatus") {
+               my ($name,$gene_id,$scaffold,$temp) = split(/\|/, $gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Mus_musculus") {
+               my ($protein,$type,$chrom,$gene_id,$transcript) = split(/\s/,$gene_header);
+               $gene_id =~ s/Gene\://;
+               $gene = $gene_id;
+       }elsif ($species eq "Neurospora_crassa") {
+               my ($transcript,$gene_id,$temp) = split(/\s\|\s/,$gene_header);
+               $gene = $transcript;
+       }elsif ($species eq "Nostoc_punctiforme") {
+               my ($temp,$transcript,$source,$gene_id,$func) = split(/\|/, $gene_header);
                $gene = $gene_id;
        }elsif ($species eq "Oryza_sativa") {
                my ($isomer,$temp,$type) = split(/\|/,$gene_header);
                $gene = $isomer;
-       }elsif ($species eq "Physcomitrella") {
-               my ($name,$locus_id,$chrom_id,$prot_id) = split(/\|/,$gene_header);
-               $gene = $prot_id; #???
-       }elsif ($species eq "Poplar") {
-               my ($name,$locus_id,$chrom_id,$prot_id) = split(/\|/,$gene_header);
+       }elsif ($species eq "Pediculus_humanus") {
+               my ($source,$gene_pa,$func,$unknown,$gene_id) = split(/\|/, $gene_header);
+               $gene_id =~ s/gene\://;
+               $gene = $gene_id;
+       }elsif ($species eq "Physcomitrella_patens") {
+               my ($name,$locus_id,$scaffold,$prot_id) = split(/\|/,$gene_header);
                $gene = $prot_id; #???
-       }elsif ($species eq "P_persica") {
+       }elsif ($species eq "Populus_trichocarpa") {
+               my ($name,$gene_id,$scaffold,$prot_id) = split(/\|/,$gene_header);
+               $gene = $gene_id; #???
+       }elsif ($species eq "Prunus_persica") {
                my ($gene_id,$temp) = split(/\s/,$gene_header);
                $gene = $gene_id;
-       }elsif ($species eq "Sacc_cerevisiae") {
-               my ($gene_id,$temp) = split(/\s/,$gene_header);
+       }elsif ($species eq "Rattus_norvegicus") {
+               my ($gene_id,$transcript,$chrom) = split(/\|/, $gene_header);
                $gene = $gene_id;
-       }elsif ($species eq "Schizo_pombe") {
-               my ($gene_id,$temp) = split(/\s/,$gene_header);
+       }elsif ($species eq "Rhizopus_oryzae") {
+               my ($gene_id,$func) = split(/\s\|\s/, $gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Ricinus_communis") {
+               my ($name,$transcript,$gene_id,$temp) = split(/\|/, $gene_header);
+               $gene = $transcript;
+       }elsif ($species eq "Saccharomyces_cerevisiae") {
+               my ($unknown,$gene_name,$gene_id,$temp) = split(/\s/,$gene_header);
+               $gene_id =~ s/SGDID\://;
+               $gene = $gene_id;
+       }elsif ($species eq "Schizosaccharomyces_pombe") {
+               my ($gene_id,$gene_name,$unknown,$func,$name,$chrom,$temp) = split(/\s/,$gene_header);
                $gene = $gene_id;
-       }elsif ($species eq "Selaginella") {
+       }elsif ($species eq "Selaginella_moellendorffii") {
                my ($name,$locus_id,$chrom_id,$prot_id) = split(/\|/,$gene_header);
                $gene = $prot_id; #???
-       }elsif ($species eq "Sorghum") {
-               my ($name,$locus_id,$scaff_id,$prot_id) = split(/\|/,$gene_header);
-               $gene = $prot_id; #???
-       }elsif ($species eq "Synechocystis") {
-               my ($gene_id,$type,$temp) = split(" ",$gene_header);
+       }elsif ($species eq "Sorghum_bicolor") {
+               my ($name,$gene_id,$chrom,$prot_id) = split(/\|/,$gene_header);
+               $gene = $gene_id; #???
+       }elsif ($species eq "Synechocystis_pcc6803") {
+               my ($gene_id,$temp) = split(" ",$gene_header);
                $gene = $gene_id; #???
-       }elsif ($species eq "Vitis") {
+       }elsif ($species eq "Trichodesmium_erythraeum") {
+               my ($temp,$unknown,$source,$gene_id,$func) = split(/\|/, $gene_header);
+               $gene = $gene_id;
+       }elsif ($species eq "Vitis_vinifera") {
                my ($name,$gene_id,$chrom_id,$id) = split(/\|/,$gene_header);
                $gene = $gene_id; #???
+       }elsif ($species eq "Zea_mays") {
+               my ($name,$gene_id,$unknown,$transcript) = split(/\|/, $gene_header);
+               $gene = $transcript
        }else {
                die "Error: Gene id can not be found for species $species!";
        }