From: elserj <elserj@localhost>
Date: Wed, 30 Sep 2009 20:36:17 +0000 (+0000)
Subject: Added subroutine to only add the gene or locus id to db instead of entire gene header... 
X-Git-Url: http://gitweb.planteome.org/?a=commitdiff_plain;h=0a4fadf5cedc049417c6446eecb32c44368306b1;p=old-jaiswallab-svn%2F.git

Added subroutine to only add the gene or locus id to db instead of entire gene header.  Still might not be right for some species.

svn path=/; revision=10
---

diff --git a/interactome_scripts/inparanoid_output_parse.pl b/interactome_scripts/inparanoid_output_parse.pl
index 5864f8c..3528d97 100755
--- a/interactome_scripts/inparanoid_output_parse.pl
+++ b/interactome_scripts/inparanoid_output_parse.pl
@@ -104,7 +104,8 @@ foreach my $input_file (@files) {
 			$id = $id_hash{$clust_id};
 		}
 	
-		$sth->execute($id, $bit_score, $species, $score, $gene_header);
+		my $gene = find_gene($gene_header,$species);
+		$sth->execute($id, $bit_score, $species, $score, $gene);
 	
 		$clust_id_prev = $clust_id;
 
@@ -165,4 +166,67 @@ sub find_species {
 	return $temp;
 }
 
+sub find_gene {
+	my $gene_header = $_[0];
+	my $species = $_[1];
+	my $gene;
+	if ($species eq "Ath") {
+		my ($name,$gene_id,$chrom,$isomer) = split("\|", $gene_header);
+		$gene = $isomer;
+	}elsif ($species eq "Brachy") {
+		$gene = $gene_header;
+	}elsif ($species eq "C_elegans") {
+		my ($gene_id,$temp) = split("\|", $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") {
+		$gene = $gene_header;
+	}elsif ($species eq "E_coli") {
+		$gene = $gene_header; #???
+	}elsif ($species eq "Soy") {
+		my ($name,$locus_id,$scaff_id,$isomer) = split("\|",$gene_header);
+		$gene = $isomer;
+	}elsif ($species eq "Human") {
+		$gene = $gene_header;
+	}elsif ($species eq "Maize") {
+		$gene = $gene_header;
+	}elsif ($species eq "Mouse") {
+		$gene = $gene_header;
+	}elsif ($species eq "Neurospora") {
+		$gene = $gene_header;
+	}elsif ($species eq "Oryza_sativa") {
+		my ($isomer,$temp,$type) = split("\|",$gene_header);
+		$gene = $isomer;
+	}elsif ($species eq "Physcomitreall") {
+		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);
+		$gene = $prot_id; #???
+	}elsif ($species eq "Sacc_cerevisiae") {
+		$gene = $gene_header;
+	}elsif ($species eq "Sacc_pombe") {
+		$gene = $gene_header;
+	}elsif ($species eq "Selaginella") {
+		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 "Synechosystis") {
+		my ($gene_id,$type,$temp) = split(" ",$gene_header);
+		$gene = $gene_id; #???
+	}elsif ($species eq "Grape") {
+		my ($name,$gene_id,$chrom_id,$id) = split("\|",$gene_header);
+		$gene = $gene_id; #???
+	}else {
+		die "Error: Gene id can not be found!";
+	}
+	return $gene;
+}
+		
+		
+