Merge pull request #227 from andrewjpage/use_full_accessory_for_tree

Provide full accessory for building binary tree
sanger-pathogens · Jan 20, 2016 · cdc517b · cdc517b
2 parents f7770a6 + 58056e4
commit cdc517b
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 58 deletions.
diff --git a/dist.ini b/dist.ini
@@ -1,5 +1,5 @@
 name    = Bio-Roary
-version = 3.5.7
+version = 3.5.8
 author  = Andrew J. Page <ap13@sanger.ac.uk>
 license = GPL_3
 copyright_holder = Wellcome Trust Sanger Institute

diff --git a/install_dependencies.sh b/install_dependencies.sh
@@ -36,7 +36,7 @@ FASTTREE_VERSION="2.1.8"
 FASTTREE_DOWNLOAD_FILENAME="FastTree-${FASTTREE_VERSION}.c"
 FASTTREE_URL="http://microbesonline.org/fasttree/FastTree-${FASTTREE_VERSION}.c"
 
-MAFFT_VERSION="7.221"
+MAFFT_VERSION="7.271"
 MAFFT_DOWNLOAD_FILENAME="mafft-${MAFFT_VERSION}-without-extensions-src.tgz"
 MAFFT_URL="http://mafft.cbrc.jp/alignment/software/${MAFFT_DOWNLOAD_FILENAME}"
 

diff --git a/lib/Bio/Roary/AccessoryBinaryFasta.pm b/lib/Bio/Roary/AccessoryBinaryFasta.pm
@@ -20,16 +20,13 @@ use Bio::Roary::Exceptions;
 use Bio::SeqIO;
 use File::Basename;
 
-has 'input_files'            => ( is => 'ro', isa => 'ArrayRef',                   required => 1 );
-has 'annotate_groups_obj'    => ( is => 'ro', isa => 'Bio::Roary::AnnotateGroups', required => 1 );
-has 'analyse_groups_obj'     => ( is => 'ro', isa => 'Bio::Roary::AnalyseGroups',  required => 1 );
-has 'output_filename'        => ( is => 'ro', isa => 'Str',                        default  => 'accessory_binary_genes.fa' );
-has 'lower_bound_percentage' => ( is => 'ro', isa => 'Int',                        default  => 5 );
-has 'upper_bound_percentage' => ( is => 'ro', isa => 'Int',                        default  => 5 );
-has 'max_accessory_to_include' => ( is => 'ro', isa => 'Int',                      default  => 4000 );
-has 'groups_to_files'        => ( is => 'ro', isa => 'HashRef',                    lazy     => 1, builder => '_build__groups_to_files' );
-has '_lower_bound_value'     => ( is => 'ro', isa => 'Int',                        lazy     => 1, builder => '_build__lower_bound_value' );
-has '_upper_bound_value'     => ( is => 'ro', isa => 'Int',                        lazy     => 1, builder => '_build__upper_bound_value' );
+has 'input_files'         => ( is => 'ro', isa => 'ArrayRef',                   required => 1 );
+has 'annotate_groups_obj' => ( is => 'ro', isa => 'Bio::Roary::AnnotateGroups', required => 1 );
+has 'analyse_groups_obj'  => ( is => 'ro', isa => 'Bio::Roary::AnalyseGroups',  required => 1 );
+has 'output_filename'     => ( is => 'ro', isa => 'Str',                        default  => 'accessory_binary_genes.fa' );
+has 'groups_to_files'     => ( is => 'ro', isa => 'HashRef',                    lazy     => 1, builder => '_build__groups_to_files' );
+has '_lower_bound_value'  => ( is => 'ro', isa => 'Int',                        lazy     => 1, builder => '_build__lower_bound_value' );
+has '_upper_bound_value'  => ( is => 'ro', isa => 'Int',                        lazy     => 1, builder => '_build__upper_bound_value' );
 
 sub _build__groups_to_files {
     my ($self) = @_;
@@ -47,48 +44,33 @@ sub _build__groups_to_files {
     return \%groups_to_files;
 }
 
-sub _build__lower_bound_value {
-    my ($self) = @_;
-    my $num_files = @{ $self->input_files };
-    return ceil( $num_files * ( $self->lower_bound_percentage / 100 ) );
-}
-
-sub _build__upper_bound_value {
-    my ($self) = @_;
-    my $num_files = @{ $self->input_files };
-    return $num_files - ceil( $num_files * ( $self->upper_bound_percentage / 100 ) );
-}
-
 sub create_accessory_binary_fasta {
     my ($self) = @_;
     my $out_seq_io = Bio::SeqIO->new( -file => ">" . $self->output_filename, -format => 'Fasta' );
 
     for my $full_filename ( @{ $self->input_files } ) {
-        my($filename, $dirs, $suffix) = fileparse($full_filename);
-        
+        my ( $filename, $dirs, $suffix ) = fileparse($full_filename);
+
         my $output_sequence = '';
         my $sample_name     = $filename;
         $sample_name =~ s!\.gff\.proteome\.faa!!;
 
-		my $gene_count = 0;
+        my $gene_count = 0;
         for my $group ( sort keys %{ $self->groups_to_files } ) {
-			last if($gene_count > $self->max_accessory_to_include);
 
             my @files = keys %{ $self->groups_to_files->{$group} };
 
-            next if ( @files <= $self->_lower_bound_value || @files > $self->_upper_bound_value );
-
             my $group_to_file_genes = $self->groups_to_files->{$group}->{$full_filename};
             if ( defined($group_to_file_genes) && @{$group_to_file_genes} > 0 ) {
                 $output_sequence .= 'A';
             }
             else {
                 $output_sequence .= 'C';
             }
-			$gene_count++;
-			
+            $gene_count++;
+
         }
-		next if($output_sequence eq '');
+        next if ( $output_sequence eq '' );
         $out_seq_io->write_seq( Bio::Seq->new( -display_id => $sample_name, -seq => $output_sequence ) );
     }
     return 1;

diff --git a/t/Bio/Roary/AccessoryBinaryFasta.t b/t/Bio/Roary/AccessoryBinaryFasta.t
@@ -33,8 +33,6 @@ ok(
             group_3 => { 't/abc/aaa' => [1], 't/abc/bbb' => [2], 't/abc/ccc' => [3] },
             group_4 => { 't/abc/aaa' => [1], 't/abc/bbb' => [2], 't/abc/ccc' => [3], 't/abc/ddd' => [4] },
         },
-		_lower_bound_value  => 0,
-		_upper_bound_value  => 4,
 		annotate_groups_obj => $dummy_annotate_groups,
 		analyse_groups_obj  => $dummy_analyse_groups
     ),
@@ -46,27 +44,4 @@ ok( $obj->create_accessory_binary_fasta(), 'create output file' );
 compare_ok( 'accessory_binary_genes.fa', 't/data/expected_accessory_binary_genes.fa','binary accessory fasta file created');
 
 
-ok(
-    $obj = Bio::Roary::AccessoryBinaryFasta->new(
-        input_files => [ 'aaa', 'bbb', 'ccc', 'ddd' ],
-        groups_to_files => 
-		{
-            group_1 => { 'aaa' => [1] },
-            group_2 => { 'aaa' => [1], 'bbb' => [2] },
-            group_3 => { 'aaa' => [1], 'bbb' => [2], 'ccc' => [3] },
-            group_4 => { 'aaa' => [1], 'bbb' => [2], 'ccc' => [3], 'ddd' => [4] },
-        },
-		annotate_groups_obj => $dummy_annotate_groups,
-		analyse_groups_obj  => $dummy_analyse_groups
-    ),
-    'initialise accessory binary fasta file bounded'
-);
-
-is($obj->_lower_bound_value, 1, 'lower bound value');
-is($obj->_upper_bound_value, 3, 'upper bound value');
-ok( $obj->create_accessory_binary_fasta(), 'create output file bounded' );
-
-compare_ok( 'accessory_binary_genes.fa', 't/data/expected_accessory_binary_genes_bounded.fa','binary accessory fasta file created bounded');
-
-
 done_testing();