From: elserj <elserj@localhost>
Date: Thu, 7 May 2020 21:44:08 +0000 (+0000)
Subject: Fix the species printing in output to not have underscore "_"
X-Git-Url: http://gitweb.planteome.org/?a=commitdiff_plain;h=63c275a4f4b97276aa17b8ff30e94c69dfeb29a8;p=old-jaiswallab-svn%2F.git

Fix the species printing in output to not have underscore "_"

svn path=/; revision=679
---

diff --git a/interactome_scripts/cafe_from_inparanoid.pl b/interactome_scripts/cafe_from_inparanoid.pl
index 5c2ab45..9ccca41 100755
--- a/interactome_scripts/cafe_from_inparanoid.pl
+++ b/interactome_scripts/cafe_from_inparanoid.pl
@@ -25,6 +25,8 @@ my $dbh = DbiFloret::dbconnect;
 my $in_file = $ARGV[0];
 my $out_file = $ARGV[1];
 
+my $max_genes = 100;
+
 # Get list of species to check
 my @species_list;
 
@@ -71,7 +73,9 @@ open(OUTFILE, ">$out_file");
 print OUTFILE "Description\tID";
 # Print species list to output file
 for my $species (@species_list) {
-		print OUTFILE "\t$species";
+	my $species_sub = $species;
+	$species_sub =~ s/_//;
+	print OUTFILE "\t$species_sub";
 }
 print OUTFILE "\n";
 
@@ -98,13 +102,17 @@ for my $super_id (sort {$a <=> $b} keys %cluster_hash) {
 		# drop any lines (clusters) that only have one species in them
 		my @line_array = split("\t", $mid_line);
 		my @match_array;
+		my $max_over = 1;
 		foreach my $elem (@line_array) {
 				if($elem != 0) {
 						push(@match_array, $elem);
 				}
+				if($elem >=  $max_genes) {
+					$max_over = 0;
+				}
 		}
 		my $match_count = @match_array;
-		if ($match_count != 1) {
+		if ($match_count != 1 && $max_over) {
 				print OUTFILE "$super_id\t$super_id\t$mid_line\n";
 		}
 }