From: elserj <elserj@localhost>
Date: Tue, 16 Dec 2014 23:46:49 +0000 (+0000)
Subject: Scripts that can be used together to run inParanoid as an SGE batch array style job... 
X-Git-Url: http://gitweb.planteome.org/?a=commitdiff_plain;h=7a87356d6708ad97679be4475071cee3f804e597;p=old-jaiswallab-svn%2F.git

Scripts that can be used together to run inParanoid as an SGE batch array style job, also removes files after completion except for sqltable files which is moved to sqltables directory

svn path=/; revision=598
---

diff --git a/interactome_scripts/create_fasta_list.sh b/interactome_scripts/create_fasta_list.sh
new file mode 100755
index 0000000..08f5988
--- /dev/null
+++ b/interactome_scripts/create_fasta_list.sh
@@ -0,0 +1,18 @@
+#!/bin/bash
+
+outfile="fasta_job_list.txt"
+
+counter=1
+
+lsarray=(`ls *.fa`)
+len=${#lsarray[@]}
+
+for (( i=0; i<$len; i++))
+do
+        for (( j=$i+1; j<$len; j++ ))
+        do
+		echo "$counter	${lsarray[$i]}	${lsarray[$j]}" >> $outfile
+		((counter++))
+	done
+done
+
diff --git a/interactome_scripts/run_array_job_by_id.pl b/interactome_scripts/run_array_job_by_id.pl
new file mode 100755
index 0000000..4340cc4
--- /dev/null
+++ b/interactome_scripts/run_array_job_by_id.pl
@@ -0,0 +1,31 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+
+
+if($#ARGV != 1) {
+	print "usage: run_array_job_by_id.pl job_id id_mapping_file\n";
+	exit;
+}
+
+my $id = $ARGV[0];
+my $mappingFile = $ARGV[1];
+
+my %map_hash;
+
+open(MAPFILE, "$mappingFile");
+while(<MAPFILE>) {
+		my $line = $_;
+		chomp $line;
+		my ($map_id, $species_1, $species_2) = split(/\s+/, $line);
+		$map_hash{$map_id} = "$species_1\t$species_2";
+}
+
+close(MAPFILE);
+
+my ($species_1, $species_2) = split(/\s+/, $map_hash{$id});
+
+system "run_inparanoid_single.sh $species_1 $species_2\n";
+
+
diff --git a/interactome_scripts/run_inparanoid_single.sh b/interactome_scripts/run_inparanoid_single.sh
new file mode 100755
index 0000000..6b32d1b
--- /dev/null
+++ b/interactome_scripts/run_inparanoid_single.sh
@@ -0,0 +1,30 @@
+#!/bin/bash
+
+currentdir=(`pwd`)
+date=(`date +%m_%d_%y`)
+directory=/lemma/inparanoid_data/$date
+mkdir -p $directory
+
+file1=$1
+file2=$2
+
+dirname=$directory/${file1}_______${file2}
+mkdir -p $dirname
+fasta_header_fix.pl ${file1} $dirname/${file1}
+fasta_header_fix.pl ${file2} $dirname/${file2}
+cd $dirname
+cp /home/bpp/elserj/Orthology_software/inparanoid/inparanoid_fixed.pl ./
+cp /home/bpp/elserj/Orthology_software/inparanoid/blast_parser.pl ./
+cp /home/bpp/elserj/Orthology_software/inparanoid/BLOSUM62 ./
+cp /home/bpp/elserj/Orthology_software/inparanoid/InparanoidXMLCreator.pl ./
+cp /home/bpp/elserj/Orthology_software/inparanoid/seqstat.jar ./
+
+cp /lemma/inparanoid_data/07_22_14/sqltables/sqltable.Triticum_urartu_gramene_ftp_1_17_14.fa-Zea_mays_phytozome_mart_9_9_13.fa ./
+
+inparanoid_fixed.pl ${file1} ${file2}
+
+mkdir -p $directory/sqltables
+mv sqltable.${file1}-${file2} $directory/sqltables
+
+cd ..
+rm -r $dirname