view release on metacpan or  search on metacpan

lib/Bio/AutomatedAnnotation/External/Cmscan.pm  view on Meta::CPAN

package Bio::AutomatedAnnotation::External::Cmscan;
$Bio::AutomatedAnnotation::External::Cmscan::VERSION = '1.182770';
# ABSTRACT: Run and parse the output of cmscan


use Moose;
use Bio::SeqFeature::Generic;

has 'cmdb'       => ( is => 'ro', isa => 'Str', required => 1 );
has 'input_file' => ( is => 'ro', isa => 'Str', required => 1 );

has 'exec'     => ( is => 'ro', isa => 'Str',     default => 'cmscan' );
has 'version'  => ( is => 'ro', isa => 'Str',     default => '1.1' );
has 'cpus'     => ( is => 'ro', isa => 'Int',     default => 1 );
has 'evalue'   => ( is => 'ro', isa => 'Num',     default => 1E-6 );
has 'features' => ( is => 'ro', isa => 'HashRef', lazy    => 1, builder => '_build_features' );
has 'number_of_features'  => ( is => 'rw', isa => 'Int',  default => 0 );

has '_tool'        => ( is => 'ro', isa => 'Str', lazy => 1, builder => '_build__tool' );
has '_cpus'        => ( is => 'ro', isa => 'Int', lazy => 1, builder => '_build__cpus' );
has '_infernal_fh' => ( is => 'ro',  lazy => 1, builder => '_build__infernal_fh' );

sub _build__tool {
    my ($self) = @_;
    return "Infernal:" . $self->version;
}

sub _build__cpus {
    my ($self) = @_;
    return $self->cpus || 1;
}

sub _build__infernal_fh {
    my ($self) = @_;
    open(
        my $fh,
        '-|',
        join(
            ' ',
            (
                $self->exec,   '--cpu', $self->_cpus, '-E',      $self->evalue, '--tblout',
                '/dev/stdout', '-o',    '/dev/null',  '--noali', $self->cmdb,   $self->input_file
            )
        )
    );
    return $fh;
}

sub _build_features {
    my ($self) = @_;
    my %features;
    
    my $number_of_features = 0;
    my $fh = $self->_infernal_fh;
    while ( <$fh> ) {
        next if (/\#/);
        my @x = split ' ';    # magic Perl whitespace splitter
        next if ( @x < 16 );
        next unless $x[1] =~ m/^RF\d/;

        # The start coord is always the lowest
        my $sequence_id = $x[2];
        my $start_coords   = $x[7];
        my $end_coords     = $x[8];
        my $current_strand = $x[9] eq '-' ? -1 : +1;

        if ( $start_coords > $end_coords ) {
            my $tmp_coords = $end_coords;
            $end_coords     = $start_coords;
            $start_coords   = $tmp_coords;
            $current_strand = -1;
        }

        push @{ $features{$sequence_id} },
          Bio::SeqFeature::Generic->new(
            -primary => 'ncRNA',