speedup searching fastas

sanger-pathogens · Apr 26, 2013 · aa9866e · aa9866e
1 parent a863be5
commit aa9866e
Show file tree

Hide file tree

Showing 4 changed files with 8 additions and 4 deletions.
diff --git a/dist.ini b/dist.ini
@@ -16,6 +16,7 @@ requires = makeblastdb
 requires = cd-hit
 requires = mcl
 requires = mcxdeblast
+requires = fasta_grep
 
 [@Basic]
 [PruneCruft]

diff --git a/lib/Bio/PanGenome/AnalyseGroups.pm b/lib/Bio/PanGenome/AnalyseGroups.pm
@@ -17,7 +17,6 @@ Take in a groups file and the original FASTA files and create plots and stats
 =cut
 
 use Moose;
-use Bio::SeqIO;
 use Bio::PanGenome::Exceptions;
 use Bio::PanGenome::Plot::FreqOfGenes;
 

diff --git a/lib/Bio/PanGenome/Output/GroupMultifasta.pm b/lib/Bio/PanGenome/Output/GroupMultifasta.pm
@@ -43,7 +43,8 @@ sub _build__genes {
 sub _lookup_sequence {
     my ( $self, $gene, $filename ) = @_;
 
-    my $fasta_obj = Bio::SeqIO->new( -file => $filename, -format => 'Fasta' );
+    open(my $fh, '-|', 'fasta_grep -f '.$filename. ' '.$gene);
+    my $fasta_obj = Bio::SeqIO->new( -fh => $fh, -format => 'Fasta' );
     while ( my $seq = $fasta_obj->next_seq() ) {
         next unless ( $seq->display_id eq $gene );
         return $seq;

diff --git a/lib/Bio/PanGenome/Output/OneGenePerGroupFasta.pm b/lib/Bio/PanGenome/Output/OneGenePerGroupFasta.pm
@@ -45,14 +45,17 @@ sub _lookup_sequences {
     {
       $gene_queue{$gene}++;
     }
+    my $gene_params = join(' ', @{$genes});
 
-
-    my $fasta_obj = Bio::SeqIO->new( -file => $filename, -format => 'Fasta' );
+    open(my $fh, '-|', 'fasta_grep -f '.$filename. ' '.$gene_params);
+    my $fasta_obj = Bio::SeqIO->new( -fh => $fh, -format => 'Fasta' );
     while ( my $seq = $fasta_obj->next_seq() ) {
+        last unless(%gene_queue);
         for my $gene ( keys %gene_queue)
         {
           next unless ( $seq->display_id eq $gene );
           push(@sequences, $seq);
+          delete($gene_queue{$gene});
         }
     }
     return \@sequences;