view release on metacpan or  search on metacpan

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

package Bio::MUST::Core::Taxonomy;
# ABSTRACT: NCBI Taxonomy one-stop shop
# CONTRIBUTOR: Loic MEUNIER <loic.meunier@doct.uliege.be>
# CONTRIBUTOR: Mick VAN VLIERBERGHE <mvanvlierberghe@doct.uliege.be>
$Bio::MUST::Core::Taxonomy::VERSION = '0.252040';
use Moose;
use namespace::autoclean;

use autodie;
use feature qw(say);

use Smart::Comments '###';

use MooseX::Storage;
with Storage('io' => 'StorableFile');

use Moose::Util::TypeConstraints;

use Algorithm::NeedlemanWunsch;
use Carp;
use Const::Fast;
use File::Basename;
use File::Find::Rule;
use IPC::System::Simple qw(system);
use List::AllUtils 0.12
    qw(first firstidx uniq each_array mesh count_by max_by);
use LWP::Simple qw(get getstore);
use Path::Class qw(dir file);
use POSIX;
use Scalar::Util qw(looks_like_number);
use Try::Tiny;
use Try::Tiny::Warnings;
use XML::Bare;

use Bio::LITE::Taxonomy::NCBI::Gi2taxid qw(new_dict);
use Bio::Phylo::IO qw(parse);

use Bio::MUST::Core::Types;
use Bio::MUST::Core::Constants qw(:ncbi :files);
use Bio::MUST::Core::Utils qw(change_suffix);
use aliased 'Bio::MUST::Core::SeqId';
use aliased 'Bio::MUST::Core::IdList';
use aliased 'Bio::MUST::Core::IdMapper';
use aliased 'Bio::MUST::Core::Tree';
use aliased 'Bio::MUST::Core::Taxonomy::MooseNCBI';
use aliased 'Bio::MUST::Core::Taxonomy::Filter';
use aliased 'Bio::MUST::Core::Taxonomy::Criterion';
use aliased 'Bio::MUST::Core::Taxonomy::Category';
use aliased 'Bio::MUST::Core::Taxonomy::Classifier';
use aliased 'Bio::MUST::Core::Taxonomy::Labeler';
use aliased 'Bio::MUST::Core::Taxonomy::ColorScheme';


# public path to NCBI Taxonomy dump directory
has 'tax_dir' => (
    traits   => ['DoNotSerialize'],
    is       => 'ro',
    isa      => 'Bio::MUST::Core::Types::Dir',
    required => 1,
    coerce   => 1,
);


# Note: init_arg => undef had to be removed to allow proper serialization
# this is needed because MooseX::Storage has never accepted my patches
# see https://rt.cpan.org/Public/Bug/Display.html?id=65733

has '_ncbi_tax' => (
    is       => 'ro',
    isa      => 'Bio::MUST::Core::Taxonomy::MooseNCBI',
    lazy     => 1,
    builder  => '_build_ncbi_tax',
    handles  => qr{get_\w+}xms, # expose Bio::LITE::Taxonomy accessor methods
);

                                # Note: this is related to (yet different from)
has '_gi_mapper' => (           # the nearly homonymous 'gi_mapper' method
    traits   => ['DoNotSerialize'],
    is       => 'ro',
    isa      => 'Bio::LITE::Taxonomy::NCBI::Gi2taxid',
    lazy     => 1,
    builder  => '_build_gi_mapper',
    handles  => {
        get_taxid_from_gi => 'get_taxid',
    },
);


has '_is_deleted' => (
    traits   => ['Hash'],
    is       => 'ro',
    isa      => 'HashRef[Bool]',