From 206f2710761fbf6fcecf8291eaa8e6b02650fc13 Mon Sep 17 00:00:00 2001
From: elserj <elserj@localhost>
Date: Wed, 14 Apr 2010 20:20:04 +0000
Subject: [PATCH] Program to query inparanoid database and retrieve orthologs
 and paralogs

svn path=/; revision=15
---
 interactome_scripts/find_ortho_inpara.pl | 74 ++++++++++++++++++++++++
 1 file changed, 74 insertions(+)
 create mode 100755 interactome_scripts/find_ortho_inpara.pl

diff --git a/interactome_scripts/find_ortho_inpara.pl b/interactome_scripts/find_ortho_inpara.pl
new file mode 100755
index 0000000..51d9504
--- /dev/null
+++ b/interactome_scripts/find_ortho_inpara.pl
@@ -0,0 +1,74 @@
+#!/usr/bin/perl
+
+#####################################################################
+#    Written by Justin Elser 4/14/10                                #
+#                                                                   #
+#    This program takes an input file with a list of genes and      #
+#      finds the orthologs (and paralogs if the score in sth_get_ortho #
+#      is changed).                                                 #
+#                                                                   #
+#####################################################################
+
+use strict;
+use warnings;
+
+use lib "$ENV{HOME}/scripts/jaiswallab/interactome_scripts";
+
+use DbiFloret;
+
+my $dbh = DbiFloret::dbconnect;
+
+#my @species_array = ("Ath", "Oryza_sativa", "Sorghum", "Maize", "Glycine");
+my @species_array = ("Ath", "Maize");
+my $spec_array_size = @species_array;
+
+# read in list of genes from csv file given as argument
+my $in_file = $ARGV[0];
+
+open (in_file, "$in_file");
+
+my @in_gene_array;
+
+while(<in_file>) {
+	my $in_gene = $_;
+	chomp $in_gene;	
+	$in_gene .= ".1"; # add the suffix back so that the gene matches the db
+	push(@in_gene_array, $in_gene);
+	
+}
+close(in_file);
+
+open(out_file, ">flower_ortho.txt");
+
+for (my $i = 1; $i<$spec_array_size; $i++) {
+	
+	my $table = $species_array[0] . "_" . $species_array[$i];
+	print out_file "$species_array[$i]\t$table\n";
+	
+	# set up the db query statement
+	my $sth_get_id = $dbh->prepare("select id from $table where gene = ?");
+	
+	# set up db query statement to get the orthologs using the cluster id
+	my $sth_get_ortho = $dbh->prepare("select gene from $table where id = ? and species = '$species_array[$i]' and score = '1'");
+	
+	foreach my $gene (@in_gene_array) {
+		
+		my $rv1 = $sth_get_id->execute($gene);
+		if (!$rv1) {
+			next;
+		}
+		
+		while (my $id = $sth_get_id->fetchrow_array()) {
+			my $rv2 = $sth_get_ortho->execute($id);
+			if (!$rv2) {
+				next;
+			}
+			
+			while (my $ortho = $sth_get_ortho->fetchrow_array()) {
+				print out_file "$gene\t$ortho\n";
+			}
+		}
+	}
+}
+
+close(out_file);
-- 
2.34.1