Added species hash list to check against and will only fix lines that match

svn path=/; revision=612

diff --git a/interactome_scripts/fix_sqltable_files.pl b/interactome_scripts/fix_sqltable_files.pl
index e83ad65e94d5295e57dfcdf43303a42a26be8e98..61c5680ebaf456135e7415a84695edbe8a30f377 100755 (executable)
--- a/interactome_scripts/fix_sqltable_files.pl
+++ b/interactome_scripts/fix_sqltable_files.pl
@@ -15,14 +15,28 @@ my $out_file = $ARGV[1];
 open(INFILE, "$in_file");
 open(OUTFILE, ">$out_file");
 
+my %species_hash;
+$species_hash{"Cannabis_sativa.Purple.Kush"} = 1;
+$species_hash{"Phyllostachys_heterocycla"} = 1;
+$species_hash{"Theobroma_cacao"} = 1;
+$species_hash{"Triticum_monococcum.DV92"} = 1;
+$species_hash{"Triticum_monococcum.G3116"} = 1;
+
+
+
+
 while(<INFILE>) {
        my $line = $_;
        chomp $line;
        
        my ($clust_id, $bit_score, $file, $score, $gene) = split(/\t/, $line);
        my $species = find_species($file);
-       my $new_gene = find_gene($gene, $species);
-       print OUTFILE "$clust_id\t$bit_score\t$file\t$score\t$new_gene\n";
+       if(defined($species_hash{$species})) {
+                       my $new_gene = find_gene($gene, $species);
+                       print OUTFILE "$clust_id\t$bit_score\t$file\t$score\t$new_gene\n";
+       }else{
+                       print OUTFILE "$clust_id\t$bit_score\t$file\t$score\t$gene\n";
+       }
 }
 close(INFILE);
 close(OUTFILE);