view release on metacpan or  search on metacpan

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

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

# use Smart::Comments;

use Const::Fast;
use List::AllUtils qw(indexes mesh partition_by pairmap);

use Bio::MUST::Core::Types;
use aliased 'Bio::MUST::Core::IdList';
use aliased 'Bio::MUST::Core::SeqMask';


has 'categories' => (
    traits   => ['Array'],
    is       => 'ro',
    isa      => 'ArrayRef[Bio::MUST::Core::Taxonomy::Category]',
    required => 1,
    handles  => {
        all_categories => 'elements',
    },
);



sub all_labels {
    my $self = shift;
    return map { $_->label } $self->all_categories;
}



sub classify {
    my $self     = shift;
    my $listable = shift;

    # loop through cats and return the first one matching input
    # this means that the cat order may affect the classification
    for my $cat ($self->all_categories) {
        return $cat->label if $cat->matches($listable);
    }

    # return undef if no suitable cat
    return;
}


# "magic" name used when a pattern has no category
const my $NOCAT => '_NOCAT_';

# TODO: come with better name for method?
# TODO: provide a shortcut if only one cat?

sub tax_masks {
    my $self = shift;
    my $ali  = shift;

    # TODO: profile and optimize as ideal_mask ?!?

    my $width = $ali->width;
    my $regex = $ali->gapmiss_regex;