view release on metacpan or  search on metacpan

lib/Bio/Pipeline/Comparison/Report/ParseVCFCompare.pm  view on Meta::CPAN

package Bio::Pipeline::Comparison::Report::ParseVCFCompare;

# ABSTRACT: Take in the output of VCF compare and return details about intersection of variants.


use Moose;
use Try::Tiny;
use Bio::Pipeline::Comparison::Types;
use Bio::Pipeline::Comparison::Exceptions;

has 'known_variant_filename'    => ( is => 'rw', isa  => 'Str', required => 1 );
has 'observed_variant_filename' => ( is => 'rw', isa  => 'Str', required => 1 );

has 'vcf_compare_exec'          => ( is => 'ro', isa  => 'Bio::Pipeline::Comparison::Executable', default => 'vcf-compare' );

has '_venn_diagram_regex'       => ( is => 'ro', isa  => 'Str', default => '^VN\t(\d+)\t([^\s]+)\s\(([\d\.]+)%\)(\t([^\s]+)\s\(([\d\.]+)%\))?$' );
has '_vcf_compare_fh'           => ( is => 'ro', lazy => 1, builder => '_build__vcf_compare_fh' );
has '_raw_venn_diagram_results' => ( is => 'ro', isa  => 'ArrayRef', lazy => 1, builder => '_build__raw_venn_diagram_results' );

sub number_of_false_positives
{
  my ($self) = @_;
  if(@{$self->_raw_venn_diagram_results} == 1)
  {
    return 0;
  }

  return $self->_number_of_uniques_for_filename($self->observed_variant_filename);
}

sub number_of_false_negatives
{
  my ($self) = @_;
  if(@{$self->_raw_venn_diagram_results} == 1)
  {
    return 0;
  }

  return $self->_number_of_uniques_for_filename($self->known_variant_filename);
}


sub number_of_known_variants
{
  my ($self) = @_;
  $self->_number_of_variants($self->known_variant_filename);
}

sub number_of_observed_variants
{
  my ($self) = @_;
  $self->_number_of_variants($self->observed_variant_filename);
}

sub _number_of_variants
{
  my ($self,$filename) = @_;
  my $number_of_variants = 0;
  for my $row_results (@{$self->_raw_venn_diagram_results})
  {
    my $number_of_files_with_overlap = @{$row_results->{files_to_percentage}};
    
    if($number_of_files_with_overlap > 0)
    {
      for(my $i = 0; $i < $number_of_files_with_overlap; $i++ )
      {
        if(defined($row_results->{files_to_percentage}->[$1]->{file_name})