Merge pull request #139 from andrewjpage/missing_id

Use locus tag when ID is missing from GFF

dist.ini

@@ -1,5 +1,5 @@
 name    = Bio-Roary
-version = 2.3.0
+version = 2.3.1
 author  = Andrew J. Page <ap13@sanger.ac.uk>
 license = GPL_3
 copyright_holder = Wellcome Trust Sanger Institute

lib/Bio/Roary.pm

@@ -10,6 +10,7 @@ Create a pan genome
 
 use Moose;
 use File::Copy;
+use Bio::Perl;
 use Bio::Roary::ParallelAllAgainstAllBlast;
 use Bio::Roary::CombinedProteome;
 use Bio::Roary::External::Cdhit;

lib/Bio/Roary/BedFromGFFRole.pm

@@ -34,21 +34,37 @@ sub _create_bed_file_from_gff {
         next if !( $feature->primary_tag =~ /$tags_regex/ );
 
         # Must have an ID tag
-        next unless ( $feature->has_tag('ID') );
+        my $gene_id = $self->_get_feature_id($feature);
+        next unless($gene_id);
 
         #filter out small genes
         next if ( ( $feature->end - $feature->start ) < $self->min_gene_size_in_nucleotides );
 
-        my ( $gene_id, @junk ) = $feature->get_tag_values('ID');
-        $gene_id =~ s!["']!!g;
-        next if ( $gene_id eq "" );
-
         my $strand = ($feature->strand > 0)? '+':'-' ;
         print {$bed_fh} join( "\t", ( $feature->seq_id, $feature->start -1, $feature->end, $gene_id, 1, $strand ) ) . "\n";
     }
     $gffio->close();
 }
 
-
+sub _get_feature_id
+{
+    my ($self, $feature) = @_;
+    my ( $gene_id, @junk ) ;
+    if ( $feature->has_tag('ID') )
+    {
+         ( $gene_id, @junk ) = $feature->get_tag_values('ID');
+    }
+    elsif($feature->has_tag('locus_tag'))
+    {
+        ( $gene_id, @junk ) = $feature->get_tag_values('locus_tag');
+    }
+    else
+    {
+        return undef;
+    }
+    $gene_id =~ s!["']!!g;
+    return undef if ( $gene_id eq "" );
+    return $gene_id ;
+}
 
 1;

lib/Bio/Roary/GeneNamesFromGFF.pm

@@ -26,42 +26,46 @@ has 'ids_to_product' => ( is => 'rw', isa => 'HashRef', default => sub { {} } );
 sub _build_ids_to_gene_name {
     my ($self) = @_;
     my %id_to_gene_name;
-
-    open( my $fh, '-|', $self->_gff_fh_input_string ) or die "Couldnt open GFF file";
-    while(<$fh>)
-    {
-      chomp;
-      my $line = $_;   
-      my $id_name;
-      if($line =~/ID=["']?([^;"']+)["']?;?/i)
-      {
-        $id_name = $1;
-        $id_name =~ s!"!!g;
-      }
-      else
-      {
-        next;
-      }
-
-      if($line =~/gene=["']?([^;"']+)["']?;?/i)
-      {
-          my $gene_name = $1;
-          $gene_name =~ s!"!!g;
-          next if ( $gene_name eq "" );
-          $id_to_gene_name{$id_name} = $gene_name;
-      }
-
-      if($line =~/product=["']?([^;,"']+)[,"']?;?/i)
-      {
-              my $product = $1;
-              $self->ids_to_product->{$id_name} = $product;
-      }
-
+
+    my $gffio = Bio::Tools::GFF->new( -file => $self->gff_file, -gff_version => 3 );
+    while ( my $feature = $gffio->next_feature() ) {
+        my $gene_id = $self->_get_feature_id($feature);
+        next unless ($gene_id);
+
+        if ( $feature->has_tag('gene') ) {
+            my ( $gene_name, @junk ) = $feature->get_tag_values('gene');
+            $gene_name =~ s!"!!g;
+            if ( $gene_name ne "" ) {
+                $id_to_gene_name{$gene_id} = $gene_name;
+            }
+        }
+        if ( $feature->has_tag('product') ) {
+            my ( $product, @junk ) = $feature->get_tag_values('product');
+            $self->ids_to_product->{$gene_id} = $product;
+        }
+
     }
-    close($fh);
+
     return \%id_to_gene_name;
 }
 
+sub _get_feature_id {
+    my ( $self, $feature ) = @_;
+    my ( $gene_id, @junk );
+    if ( $feature->has_tag('ID') ) {
+        ( $gene_id, @junk ) = $feature->get_tag_values('ID');
+    }
+    elsif ( $feature->has_tag('locus_tag') ) {
+        ( $gene_id, @junk ) = $feature->get_tag_values('locus_tag');
+    }
+    else {
+        return undef;
+    }
+    $gene_id =~ s!["']!!g;
+    return undef if ( $gene_id eq "" );
+    return $gene_id;
+}
+
 no Moose;
 __PACKAGE__->meta->make_immutable;
 

t/Bio/Roary/CommandLine/Roary.t

@@ -70,6 +70,16 @@ ok((-e 'query_2.gff.proteome.faa'),'Check protein query_2.gff.proteome.faa is no
 ok((-e 'query_5.gff.proteome.faa'),'Check protein query_5.gff.proteome.faa is not cleaned up'); 
 
 cleanup_files();
+
+%scripts_and_expected_files = (
+'-j Local --dont_delete_files t/data/locus_tag_gffs/query_1.gff t/data/locus_tag_gffs/query_2.gff t/data/locus_tag_gffs/query_3.gff ' =>
+  [  'empty_file', 't/data/empty_file'  ],
+  );
+mock_execute_script_and_check_output_sorted_groups( $script_name, \%scripts_and_expected_files, [0,6,7,8,9] );
+
+for my $filename (('query_1.gff.proteome.faa','query_2.gff.proteome.faa','query_3.gff.proteome.faa')) {
+    is( read_file($filename), read_file( 't/data/locus_tag_gffs/' . $filename . '.expected' ), "content of proteome $filename as expected" );
+}
 
 SKIP: 
 {

t/Bio/Roary/ExtractProteomeFromGFFs.t

@@ -54,4 +54,26 @@ unlink('genbank1.gff.proteome.faa');
 unlink('genbank2.gff.proteome.faa');
 unlink('genbank3.gff.proteome.faa');
 
+
+
+
+ok(
+    $plot_groups_obj = Bio::Roary::ExtractProteomeFromGFFs->new(
+        gff_files => [ 't/data/locus_tag_gffs/query_1.gff', 't/data/locus_tag_gffs/query_2.gff', 't/data/locus_tag_gffs/query_3.gff' ],
+    ),
+    'initialise object with locus tag id gff files'
+);
+@sorted_fasta_files = sort( @{ $plot_groups_obj->fasta_files() } );
+@sorted_expected_files = sort( ( 'query_1.gff.proteome.faa', 'query_2.gff.proteome.faa', 'query_3.gff.proteome.faa' ) );
+
+is_deeply( \@sorted_fasta_files, \@sorted_expected_files, 'locus tag id files created output' );
+
+for my $filename (@sorted_expected_files) {
+    is( read_file($filename), read_file( 't/data/locus_tag_gffs/' . $filename . '.expected' ), "content of proteome $filename as expected" );
+}
+
+unlink('query_1.gff.proteome.faa');
+unlink('query_2.gff.proteome.faa');
+unlink('query_3.gff.proteome.faa');
+
 done_testing();

t/Bio/Roary/GeneNamesFromGFF.t

@@ -50,4 +50,26 @@ is_deeply(
     'ids to gene names as expected again'
 );
 
+
+ok(
+    $obj = Bio::Roary::GeneNamesFromGFF->new(
+        gff_file => 't/data/locus_tag_gffs/query_1.gff'
+    ),
+    'initialise a GFF file with locus tags only'
+);
+
+is_deeply(
+    $obj->ids_to_gene_name,
+    {
+              'abc_00005' => 'speH',
+              'abc_00007' => 'argF',
+              'abc_00001' => 'different',
+              'abc_00016' => 'yfnB',
+              'abc_00008' => 'arcC1'
+            },
+    'ids to gene names with GFF file with locus tags only'
+);
+
+
+
 done_testing();