Merge pull request #256 from andrewjpage/speed_up_alignments

Speed up alignments
sanger-pathogens · Jul 1, 2016 · 684c726 · 684c726
2 parents 83faa7f + 195462f
commit 684c726
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 15 deletions.
diff --git a/lib/Bio/Roary.pm b/lib/Bio/Roary.pm
@@ -115,7 +115,7 @@ sub run {
     unlink($output_blast_results_filename) unless($self->dont_delete_files == 1);
 
     my $post_analysis = Bio::Roary::External::PostAnalysis->new(
-        job_runner                  => $self->job_runner,
+        job_runner                  => 'Local',
         cpus                        => $self->cpus,
         fasta_files                 => $self->fasta_files,
         input_files                 => $self->input_files,

diff --git a/lib/Bio/Roary/CommandLine/RoaryPostAnalysis.pm b/lib/Bio/Roary/CommandLine/RoaryPostAnalysis.pm
@@ -146,7 +146,7 @@ sub run {
       }
       else
       {
-          $job_runner_to_use = 'Local';
+          $job_runner_to_use = 'Parallel';
       }
 
       my $output_gene_files = $self->_find_input_files;

diff --git a/lib/Bio/Roary/External/GeneAlignmentFromNucleotides.pm b/lib/Bio/Roary/External/GeneAlignmentFromNucleotides.pm
@@ -31,7 +31,7 @@ has 'mafft'                       => ( is => 'ro', isa => 'Bool',     default =>
 has 'dont_delete_files'           => ( is => 'rw', isa => 'Bool',     default  => 0 );
 has 'num_input_files'             => ( is => 'ro', isa => 'Int',      required => 1);
 
-# Overload Role
+# Overload Role`
 has 'memory_in_mb' => ( is => 'rw', isa => 'Int', lazy     => 1, builder => '_build_memory_in_mb' );
 has '_min_memory_in_mb'      => ( is => 'ro', isa => 'Int', default => 1500 );
 has '_max_memory_in_mb'      => ( is => 'ro', isa => 'Int', default => 60000 );
@@ -43,15 +43,7 @@ has '_dependancy_memory_in_mb'  => ( is => 'ro', isa => 'Int', default => 15000
 sub _build__files_per_chunk
 {
     my ($self) = @_;
-    if($self->num_input_files > 1000)
-    {
-               return 5;
-    }
-    elsif($self->num_input_files > 500)
-    {
-               return 7;
-    }
-    return 10;
+    return 1;
 }
 
 sub _build_memory_in_mb {

diff --git a/lib/Bio/Roary/JobRunner/Parallel.pm b/lib/Bio/Roary/JobRunner/Parallel.pm
@@ -18,6 +18,8 @@ package Bio::Roary::JobRunner::Parallel;
 use Moose;
 use File::Temp qw/ tempfile /;
 use Log::Log4perl qw(:easy);
+use File::Slurper 'write_text';
+use File::Temp qw/ tempfile /;
 
 has 'commands_to_run' => ( is => 'ro', isa => 'ArrayRef', required => 1 );
 has 'cpus'            => ( is => 'ro', isa => 'Int',      default => 1 );
@@ -28,13 +30,17 @@ has 'memory_in_mb'    => ( is => 'rw', isa => 'Int',  default => '200' );
 sub run {
     my ($self) = @_;
 
+	  my($fh, $temp_command_filename) = tempfile();
+	  write_text($temp_command_filename, join("\n", @{ $self->commands_to_run }) );
+
     for my $command_to_run(@{ $self->commands_to_run })
     {
        $self->logger->info($command_to_run);
     }
-	open(my $fh,"|-","parallel --gnu -j ".$self->cpus) || die "GNU Parallel failed";
-	print $fh join("\n", @{ $self->commands_to_run });
-	close $fh;
+		my $parallel_command  = "parallel --gnu -j ".$self->cpus." < ".$temp_command_filename ;
+		$self->logger->info($parallel_command );
+
+		system($parallel_command);
     1;
 }