view release on metacpan or  search on metacpan

lib/Bio/MUST/Core/Taxonomy/Criterion.pm  view on Meta::CPAN

package Bio::MUST::Core::Taxonomy::Criterion;
# ABSTRACT: Helper class for multiple-criterion classifier based on taxonomy
$Bio::MUST::Core::Taxonomy::Criterion::VERSION = '0.252040';
use Moose;
use namespace::autoclean;

use List::AllUtils qw(sum count_by uniq_by);

use Bio::MUST::Core::Types;


has 'tax_filter' => (
    is       => 'ro',
    isa      => 'Bio::MUST::Core::Taxonomy::Filter',
    required => 1,
    handles  => [ qw(is_allowed) ],
);


has 'min_seq_count' => (
    is       => 'ro',
    isa      => 'Num',
    default  => 1,
);


has $_ => (
    is       => 'ro',
    isa      => 'Maybe[Num]',
    default  => undef,
) for qw(              max_seq_count
         min_org_count max_org_count
         min_copy_mean max_copy_mean
         min_seq_perc  max_seq_perc
         min_org_perc  max_org_perc
);



sub matches {           ## no critic (Subroutines::ProhibitExcessComplexity)
    my $self     = shift;
    my $listable = shift;

    # case 1: handle classification of single ids

    # this should work for:
    # - SeqId objects
    # - stringified lineages
    # - mere strings
    unless ( ref $listable && $listable->can('all_seq_ids') ) {
        # TODO: make this robust to ArrayRef[] (via coercion)
        return $self->is_allowed($listable);
    }

    # case 2: handle "true" listable objects

    # get seq_ids passing tax_filter
    my @seq_ids = $listable->all_seq_ids;
    my @targets = grep { $self->is_allowed($_) } @seq_ids;

    # return success if positively avoided taxa are indeed absent
    # Note: no need to compute percentages yet
    my $seq_n = @targets;
    unless ($seq_n) {
        return 1