Bio-Pipeline-Comparison

 view release on metacpan or  search on metacpan

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

  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})
           && $row_results->{files_to_percentage}->[$1]->{file_name} eq $filename)
        {
          $number_of_variants +=$row_results->{number_of_sites};
          last;
        }
      }
    }
  }
  return $number_of_variants;
}

sub _number_of_uniques_for_filename
{
  my ($self, $filename) = @_;
  for my $row_results (@{$self->_raw_venn_diagram_results})
  {
    if(@{$row_results->{files_to_percentage}} == 1
      && defined($row_results->{files_to_percentage}->[0]->{file_name})
      && $row_results->{files_to_percentage}->[0]->{file_name} eq $filename)
    {
      return $row_results->{number_of_sites};
    }
  }
  return 0;
}


sub _build__vcf_compare_fh
{

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

  my $vd_regex = $self->_venn_diagram_regex;
  my $fh = $self->_vcf_compare_fh;
  seek($fh, 0, 0);
  while(<$fh>)
  {
    my $line = $_;
    if( $line =~ m/$vd_regex/)
    {
      my %vd_results;
      $vd_results{number_of_sites} = $1;
      $vd_results{files_to_percentage} = [ {file_name => $2, percentage => $3} ];
      if(defined($4) && defined($5) && defined($6))
      {
        push(@{$vd_results{files_to_percentage}}, {file_name => $5, percentage => $6} );
      }
      push(@vd_rows,\%vd_results);
    }
  }
  return \@vd_rows;
}

no Moose;
__PACKAGE__->meta->make_immutable;
1;

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

}

ok((my $obj = Bio::Pipeline::Comparison::Report::ParseVCFCompare->new(
      known_variant_filename    => 't/data/no_overlap/known_no_overlap.vcf.gz', 
      observed_variant_filename => 't/data/no_overlap/observed_no_overlap.vcf.gz',
      'vcf_compare_exec'        => abs_path('bin/vcf-compare')
      )
    ), 'Initialise no overlap');
my @expected_no_overlap =  (
          {
            'files_to_percentage' => [
                {
                  'percentage' => '100.0',
                  'file_name' => 't/data/no_overlap/observed_no_overlap.vcf.gz'
                }
              ],
              'number_of_sites' => '5'
          },
          {
            'files_to_percentage' => [
              {
                'file_name' => 't/data/no_overlap/known_no_overlap.vcf.gz',
                'percentage' => '100.0',
              }
            ],
            'number_of_sites' => '5'
          }
        );
is_deeply($obj->_raw_venn_diagram_results, \@expected_no_overlap, 'Results from no overlap' );
is($obj->number_of_false_positives, 5 ,'no overlap fp');
is($obj->number_of_false_negatives, 5 ,'no overlap fn');


ok((my $obj_perfect = Bio::Pipeline::Comparison::Report::ParseVCFCompare->new(
      known_variant_filename    => 't/data/perfect/known_perfect.vcf.gz', 
      observed_variant_filename => 't/data/perfect/observed_perfect.vcf.gz',
      'vcf_compare_exec'        => abs_path('bin/vcf-compare')
      )
    ), 'Initialise all variants the same');
my @expected_perfect =  (
  {
    'files_to_percentage' => [
                               {
                                 'file_name' => 't/data/perfect/known_perfect.vcf.gz',
                                 'percentage' => '100.0'
                               },
                               {
                                 'file_name' => 't/data/perfect/observed_perfect.vcf.gz',
                                 'percentage' => '100.0'
                               }
                             ],
    'number_of_sites' => '5'
  }
);
is_deeply($obj_perfect->_raw_venn_diagram_results, \@expected_perfect, 'Results from perfect results' );
is($obj_perfect->number_of_false_positives, 0 ,'perfect fp');
is($obj_perfect->number_of_false_negatives, 0 ,'perfect fn');


ok((my $obj_fn = Bio::Pipeline::Comparison::Report::ParseVCFCompare->new(
      known_variant_filename    => 't/data/false_negatives/known_false_negatives.vcf.gz', 
      observed_variant_filename => 't/data/false_negatives/observed_false_negatives.vcf.gz',
      'vcf_compare_exec'        => abs_path('bin/vcf-compare')
      )
    ), 'Initialise variants with false negatives');
my @expected_fn =  
  (
    {
      'files_to_percentage' => [
                                 {
                                   'file_name' => 't/data/false_negatives/known_false_negatives.vcf.gz',
                                   'percentage' => '40.0'
                                 }
                               ],
      'number_of_sites' => '2'
    },
    {
      'files_to_percentage' => [
                                 {
                                   'file_name' => 't/data/false_negatives/known_false_negatives.vcf.gz',
                                   'percentage' => '60.0'
                                 },
                                 {
                                   'file_name' => 't/data/false_negatives/observed_false_negatives.vcf.gz',
                                   'percentage' => '100.0'
                                 }
                               ],
      'number_of_sites' => '3'
    }
);

is_deeply($obj_fn->_raw_venn_diagram_results, \@expected_fn, 'Results from false negatives' );
is($obj_fn->number_of_false_positives, 0 ,'only false negatives fp');
is($obj_fn->number_of_false_negatives, 2 ,'only false negatives fn');


ok((my $obj_fp = Bio::Pipeline::Comparison::Report::ParseVCFCompare->new(
      known_variant_filename    => 't/data/false_positives/known_false_positives.vcf.gz', 
      observed_variant_filename => 't/data/false_positives/observed_false_positives.vcf.gz',
      'vcf_compare_exec'        => abs_path('bin/vcf-compare')
      )
    ), 'Initialise variants with false positives');
my @expected_fp =  
  (
    {
      'files_to_percentage' => [
                                 {
                                   'file_name' => 't/data/false_positives/observed_false_positives.vcf.gz',
                                   'percentage' => '16.7'
                                 }
                               ],
      'number_of_sites' => '1'
    },
    {
      'files_to_percentage' => [
                                 {
                                   'file_name' => 't/data/false_positives/known_false_positives.vcf.gz',
                                   'percentage' => '100.0'
                                 },
                                 {
                                   'file_name' => 't/data/false_positives/observed_false_positives.vcf.gz',
                                   'percentage' => '83.3'
                                 }
                               ],
      'number_of_sites' => '5'
    }
);

is_deeply($obj_fp->_raw_venn_diagram_results, \@expected_fp, 'Results from false positives' );
is($obj_fp->number_of_false_positives, 1 ,'only false positives fp');
is($obj_fp->number_of_false_negatives, 0 ,'only false positives fn');

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

ok((my $obj_fp_and_fn = Bio::Pipeline::Comparison::Report::ParseVCFCompare->new(
      known_variant_filename    => 't/data/false_positives_and_negatives/known_false_positives_and_negatives.vcf.gz', 
      observed_variant_filename => 't/data/false_positives_and_negatives/observed_false_positives_and_negatives.vcf.gz',
      'vcf_compare_exec'        => abs_path('bin/vcf-compare')
      )
    ), 'Initialise variants with false positives and false negatives');
my @expected_fp_and_fn =  
  (
    {
       'files_to_percentage' => [
                                  {
                                    'file_name' => 't/data/false_positives_and_negatives/observed_false_positives_and_negatives.vcf.gz',
                                    'percentage' => '25.0'
                                  }
                                ],
       'number_of_sites' => '1'
     },
     {
       'files_to_percentage' => [
                                  {
                                    'file_name' => 't/data/false_positives_and_negatives/known_false_positives_and_negatives.vcf.gz',
                                    'percentage' => '25.0'
                                  }
                                ],
       'number_of_sites' => '1'
     },
     {
       'files_to_percentage' => [
                                  {
                                    'file_name' => 't/data/false_positives_and_negatives/known_false_positives_and_negatives.vcf.gz',
                                    'percentage' => '75.0'
                                  },
                                  {
                                    'file_name' => 't/data/false_positives_and_negatives/observed_false_positives_and_negatives.vcf.gz',
                                    'percentage' => '75.0'
                                  }
                                ],
       'number_of_sites' => '3'
     }
);

is_deeply($obj_fp_and_fn->_raw_venn_diagram_results, \@expected_fp_and_fn, 'Results from false positives and false negatives' );
is($obj_fp_and_fn->number_of_false_positives, 1 ,'false positives and false negatives fp');
is($obj_fp_and_fn->number_of_false_negatives, 1 ,'false positives and false negatives fn');