Merge pull request #343 from andrewjpage/core_paralogs

Optionally allow paralogs in core gene alignment

dist.ini

@@ -1,5 +1,5 @@
 name    = Bio-Roary
-version = 3.9.0
+version = 3.9.1
 author  = Andrew J. Page <[email protected]>
 license = GPL_3
 copyright_holder = Wellcome Trust Sanger Institute

lib/Bio/Roary.pm

@@ -48,6 +48,7 @@ has 'core_definition'             => ( is => 'rw', isa => 'Num',      default  =
 has 'verbose'                     => ( is => 'rw', isa => 'Bool',     default  => 0 );
 has 'mafft'                       => ( is => 'ro', isa => 'Bool',     default  => 0 );
 has 'inflation_value'             => ( is => 'rw', isa => 'Num',      default  => 1.5 );
+has 'allow_paralogs'              => ( is => 'rw', isa => 'Bool',     default => 0 );
 
 has 'output_multifasta_files' => ( is => 'ro', isa => 'Bool', default => 0 );
 
@@ -136,6 +137,7 @@ sub run {
         core_definition             => $self->core_definition,
 		verbose                     => $self->verbose,
 		mafft                       => $self->mafft,
+		allow_paralogs              => $self->allow_paralogs,
     );
     $post_analysis->run();
 

lib/Bio/Roary/CommandLine/Roary.pm

@@ -47,6 +47,7 @@ has 'dont_split_groups'       => ( is => 'rw', isa => 'Bool', default => 0 );
 has 'verbose_stats'           => ( is => 'rw', isa => 'Bool', default => 0 );
 has 'translation_table'       => ( is => 'rw', isa => 'Int',  default => 11 );
 has 'mafft'                   => ( is => 'rw', isa => 'Bool', default => 0 );
+has 'allow_paralogs'          => ( is => 'rw', isa => 'Bool', default => 0 );
 has 'group_limit'             => ( is => 'rw', isa => 'Num',  default => 50000 );
 has 'core_definition'         => ( is => 'rw', isa => 'Num',  default => 0.99 );
 has 'verbose'                 => ( is => 'rw', isa => 'Bool', default => 0 );
@@ -71,7 +72,7 @@ sub BUILD {
         $job_runner,            $makeblastdb_exec,  $mcxdeblast_exec,         $mcl_exec,      $blastp_exec,
         $apply_unknowns_filter, $cpus,              $output_multifasta_files, $verbose_stats, $translation_table,
         $run_qc,                $core_definition,   $help,                    $kraken_db,     $cmd_version,
-        $mafft,                 $output_directory,  $check_dependancies, $inflation_value,
+        $mafft,                 $output_directory,  $check_dependancies, $inflation_value, $allow_paralogs,
     );
 
     GetOptionsFromArray(
@@ -98,6 +99,7 @@ sub BUILD {
         'cd|core_definition=f'      => \$core_definition,
         'v|verbose'                 => \$verbose,
         'n|mafft'                   => \$mafft,
+		'ap|allow_paralogs'         => \$allow_paralogs,
         'k|kraken_db=s'             => \$kraken_db,
         'w|version'                 => \$cmd_version,
         'a|check_dependancies'      => \$check_dependancies,
@@ -302,7 +304,8 @@ sub run {
         core_definition         => $self->core_definition,
         verbose                 => $self->verbose,
         mafft                   => $self->mafft,
-	inflation_value         => $self->inflation_value,
+        allow_paralogs          => $self->allow_paralogs,
+	    inflation_value         => $self->inflation_value,
     );
     $pan_genome_obj->run();
 
@@ -343,11 +346,12 @@ Options: -p INT    number of threads [1]
          -r        create R plots, requires R and ggplot2
          -s        dont split paralogs
          -t INT    translation table [11]
+         -ap       allow paralogs in core alignment
          -z        dont delete intermediate files
          -v        verbose output to STDOUT
          -w        print version and exit
          -y        add gene inference information to spreadsheet, doesnt work with -e
-	 -iv STR   Change the MCL inflation value [1.5]
+         -iv STR   Change the MCL inflation value [1.5]
          -h        this help message
 
 Example: Quickly generate a core gene alignment using 8 threads

lib/Bio/Roary/CommandLine/RoaryCoreAlignment.pm

@@ -27,13 +27,14 @@ has 'spreadsheet_filename'      => ( is => 'rw', isa => 'Str', default => 'gene_
 has 'output_filename'           => ( is => 'rw', isa => 'Str', default => 'core_gene_alignment.aln' );
 has 'core_definition'           => ( is => 'rw', isa => 'Num', default => 0.99 );
 has 'dont_delete_files'         => ( is => 'rw', isa => 'Bool', default => 0 );
+has 'allow_paralogs'            => ( is => 'rw', isa => 'Bool', default => 0 );
 has '_error_message'            => ( is => 'rw', isa => 'Str' );
 has 'verbose'                   => ( is => 'rw', isa => 'Bool', default => 0 );
 
 sub BUILD {
     my ($self) = @_;
 
-    my ( $multifasta_base_directory, $spreadsheet_filename, $output_filename, $core_definition,$verbose,  $help, $mafft, $dont_delete_files );
+    my ( $multifasta_base_directory, $spreadsheet_filename, $output_filename, $core_definition,$verbose,  $help, $mafft, $allow_paralogs, $dont_delete_files );
 
     GetOptionsFromArray(
         $self->args,
@@ -42,6 +43,7 @@ sub BUILD {
         'o|output_filename=s'           => \$output_filename,
         'cd|core_definition=f'          => \$core_definition,
         'z|dont_delete_files'           => \$dont_delete_files,
+		'p|allow_paralogs'              => \$allow_paralogs,
 		'v|verbose'                     => \$verbose,
         'h|help'                        => \$help,
     );
@@ -51,6 +53,7 @@ sub BUILD {
         $self->logger->level(10000);
     }
     $self->help($help) if(defined($help));
+	$self->allow_paralogs($allow_paralogs) if(defined($allow_paralogs));
 
     if ( defined($multifasta_base_directory) && ( -d $multifasta_base_directory ) ) {
         $self->multifasta_base_directory( abs_path($multifasta_base_directory));
@@ -95,7 +98,8 @@ sub run {
 	$self->logger->info("Extract core genes from spreadsheet");
     my $core_genes_obj = Bio::Roary::ExtractCoreGenesFromSpreadsheet->new( 
         spreadsheet     => $self->spreadsheet_filename,
-        core_definition => $self->core_definition
+        core_definition => $self->core_definition,
+		allow_paralogs => $self->allow_paralogs
     );
 
 	$self->logger->info("Looking up genes in files");
@@ -130,6 +134,7 @@ Options: -o STR    output filename [core_gene_alignment.aln]
          -cd FLOAT percentage of isolates a gene must be in to be core [99]
          -m STR    directory containing gene multi-FASTAs [pan_genome_sequences]
          -s STR    gene presence and absence spreadsheet [gene_presence_absence.csv]
+         -p        allow paralogs
          -z        dont delete intermediate files
          -v        verbose output to STDOUT
          -h        this help message

lib/Bio/Roary/CommandLine/RoaryPostAnalysis.pm

@@ -41,14 +41,15 @@ has 'group_limit'                 => ( is => 'rw', isa => 'Num',  default => 500
 has 'core_definition'             => ( is => 'rw', isa => 'Num',  default => 0.99 );
 has 'verbose'                     => ( is => 'rw', isa => 'Bool', default => 0 );
 has 'mafft'                       => ( is => 'rw', isa => 'Bool', default => 0 );
+has 'allow_paralogs'              => ( is => 'rw', isa => 'Bool', default => 0 );
 
 sub BUILD {
     my ($self) = @_;
 
     my ( 
       $output_filename, $dont_create_rplots, $dont_delete_files, $dont_split_groups, $output_pan_geneome_filename, 
       $job_runner, $output_statistics_filename, $output_multifasta_files, $clusters_filename, $core_definition,
-      $fasta_files, $input_files, $verbose_stats, $translation_table, $help, $cpus,$group_limit,$verbose,$mafft
+      $fasta_files, $input_files, $verbose_stats, $translation_table, $help, $cpus,$group_limit,$verbose,$mafft, $allow_paralogs
     );
 
 
@@ -72,6 +73,7 @@ sub BUILD {
         'cd|core_definition=f'      => \$core_definition,
 		'v|verbose'                 => \$verbose,
 		'n|mafft'                   => \$mafft,
+		'q|allow_paralogs'          => \$allow_paralogs,
         'h|help'                    => \$help,
     );
 
@@ -93,6 +95,7 @@ sub BUILD {
     $self->group_limit($group_limit)                                 if ( defined($group_limit) );
     $self->core_definition( $core_definition/100 )                   if ( defined($core_definition) );
 	$self->mafft($mafft)                                             if ( defined($mafft) );
+	$self->allow_paralogs($allow_paralogs)                           if ( defined($allow_paralogs) );
     if ( defined($verbose) ) {
         $self->verbose($verbose);
         $self->logger->level(10000);
@@ -158,6 +161,7 @@ sub run {
         cpus                => $self->cpus,
 		verbose             => $self->verbose,
 		mafft               => $self->mafft,
+		allow_paralogs      => $self->allow_paralogs,
         dont_delete_files   => $self->dont_delete_files,
         num_input_files     => $#{$input_files},
       );
@@ -222,6 +226,7 @@ Options: -a        dont delete intermediate files
          -n        fast core gene alignement with MAFFT instead of PRANK
          -o STR    clusters output filename [clustered_proteins]
          -p STR    output pan genome filename [pan_genome.fa]
+         -q        allow paralogs in core alignment
          -s STR    output gene presence and absence filename [gene_presence_absence.csv]
          -t INT    translation table [11]
          -z INT    number of threads [1]

lib/Bio/Roary/External/GeneAlignmentFromNucleotides.pm

@@ -29,6 +29,7 @@ has 'translation_table'           => ( is => 'rw', isa => 'Int',      default =>
 has 'core_definition'             => ( is => 'ro', isa => 'Num',      default => 1 );
 has 'mafft'                       => ( is => 'ro', isa => 'Bool',     default => 0 );
 has 'dont_delete_files'           => ( is => 'rw', isa => 'Bool',     default  => 0 );
+has 'allow_paralogs'              => ( is => 'rw', isa => 'Bool',     default  => 0 );
 has 'num_input_files'             => ( is => 'ro', isa => 'Int',      required => 1);
 
 # Overload Role`
@@ -85,6 +86,7 @@ sub _build__core_alignment_cmd {
     my $core_cmd = "pan_genome_core_alignment";
     $core_cmd .= " -cd " . ($self->core_definition*100) if ( defined $self->core_definition );
     $core_cmd .= " --dont_delete_files " if ( defined $self->dont_delete_files  && $self->dont_delete_files == 1 );
+	$core_cmd .= " --allow_paralogs " if ( defined $self->allow_paralogs  && $self->allow_paralogs == 1 );
 
     return $core_cmd;
 }

lib/Bio/Roary/External/PostAnalysis.pm

@@ -37,6 +37,7 @@ has 'group_limit'                 => ( is => 'rw', isa => 'Num',  default  => 50
 has 'core_definition'             => ( is => 'ro', isa => 'Num',  default  => 1.0 );
 has 'verbose'                     => ( is => 'rw', isa => 'Bool', default  => 0 );
 has 'mafft'                       => ( is => 'ro', isa => 'Bool', default  => 0 );
+has 'allow_paralogs'              => ( is => 'ro', isa => 'Bool', default  => 0 );
 has '_working_directory'          => ( is => 'ro', isa => 'File::Temp::Dir', default => sub { File::Temp->newdir( DIR => getcwd, CLEANUP => 1 ); } );
 has '_gff_fofn'                   => ( is => 'ro', isa => 'Str', lazy => 1, builder => '_build__gff_fofn' );
 has '_fasta_fofn'                 => ( is => 'ro', isa => 'Str', lazy => 1, builder => '_build__fasta_fofn'  );
@@ -137,6 +138,9 @@ sub _command_to_run {
 
     my $verbose_flag = '';
     $verbose_flag = '-v' if ( defined($self->verbose) && $self->verbose == 1 );
+
+	my $allow_paralogs_flag = '';
+	$allow_paralogs_flag = '--allow_paralogs'  if ( defined($self->allow_paralogs) && $self->allow_paralogs == 1 );
 
     return join(
         " ",
@@ -156,6 +160,7 @@ sub _command_to_run {
             $verbose_stats_flag,
 			$verbose_flag,
 			$mafft_flag,
+			$allow_paralogs_flag,
             '-j', $self->job_runner,
             '--processors', $self->cpus,
             '--group_limit', $self->group_limit,