view release on metacpan or  search on metacpan

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

package Bio::MUST::Core::PostPred;
# ABSTRACT: Posterior predictive tests for sequences
$Bio::MUST::Core::PostPred::VERSION = '0.252040';
use Moose;
use namespace::autoclean;

use autodie;
use feature qw(say);

# use Smart::Comments;

use Statistics::Descriptive;
use Tie::IxHash;

use Bio::MUST::Core::Types;
use aliased 'Bio::MUST::Core::PostPred::Composition';


# private hash containing test statistics by sequence (and globally)
# Note: this hash is actually a Tie::IxHash (see factory methods)
has '_stats_for' => (
    traits   => ['Hash'],
    is       => 'ro',
    isa      => 'HashRef[ArrayRef[Num]]',
    required => 1,
    handles  => {
          all_ids => 'keys',
        stats_for => 'get',
    },
);


# private hash containing test Z-scores by sequence (and globally)
# Note: this hash is actually a Tie::IxHash (see builder)
has '_zscore_for' => (
    traits   => ['Hash'],
    is       => 'ro',
    isa      => 'HashRef[Num]',
    init_arg => undef,
    lazy     => 1,
    builder  => '_build_zscore_for',
    handles  => {
        zscore_for => 'get',
    },
);

## no critic (ProhibitUnusedPrivateSubroutines)

# TODO: switch to BUILD to compute p-values as well
# TODO: explore other Statistics:: modules on CPAN

sub _build_zscore_for {
    my $self = shift;

    tie my %zscore_for, 'Tie::IxHash';

    # loop through all ids (including $GLOBAL id if any)
    for my $id ($self->all_ids) {

        # compute std-dev from simulated stat dist
        # Note: the first value MUST BE the observed test stat
        my ($obs, @sims) = @{ $self->stats_for($id) };
        my $stat = Statistics::Descriptive::Full->new();
           $stat->add_data(@sims);
        my $mean = $stat->mean;