view release on metacpan or  search on metacpan

lib/Bio/EnsEMBL/ChainedAssemblyMapper.pm  view on Meta::CPAN

See the License for the specific language governing permissions and
limitations under the License.

=cut


=head1 CONTACT

  Please email comments or questions to the public Ensembl
  developers list at <http://lists.ensembl.org/mailman/listinfo/dev>.

  Questions may also be sent to the Ensembl help desk at
  <http://www.ensembl.org/Help/Contact>.

=cut

=head1 NAME

Bio::EnsEMBL::ChainedAssemblyMapper -
Handles mapping between two coordinate systems using the information
stored in the assembly table

=head1 SYNOPSIS

    $db   = Bio::EnsEMBL::DBSQL::DBAdaptor->new(...);
    $asma = $db->get_AssemblyMapperAdaptor();
    $csa  = $db->get_CoordSystemAdaptor();

    my $chr_cs = $cs_adaptor->fetch_by_name( 'chromosome', 'NCBI33' );
    my $cln_cs = $cs_adaptor->fetch_by_name('clone');

    $asm_mapper = $map_adaptor->fetch_by_CoordSystems( $cs1, $cs2 );

    # Map to contig coordinate system from chromosomal
    @cln_coords =
      $asm_mapper->map( 'X', 1_000_000, 2_000_000, 1, $chr_cs );

    # Map to chromosome coordinate system from contig
    @chr_coords =
      $asm_mapper->map( 'AL30421.1', 100, 10000, -1, $cln_cs );

    # List contig names for a region of chromsome
    @cln_ids = $asm_mapper->list_ids( '13', 1_000_000, 1, $chr_cs );

    # List chromosome names for a contig region
    @chr_ids =
      $asm_mapper->list_ids( 'AL30421.1', 1, 1000, -1, $cln_cs );

=head1 DESCRIPTION

The ChainedAssemblyMapper is an extension of the regular AssemblyMapper
that allows for mappings between coordinate systems that require
multi-step mapping.  For example if explicit mappings are defined
between the following coordinate systems,

  chromosome <-> contig
  contig     <-> clone

the ChainedAssemblyMapper would be able to perform implicit mapping
between the chromosome and clone coordinate systems.  This should be
transparent to the user of this module, and users should not even
realise that they are using a chained assembly mapper as opposed to a
normal assembly mapper.

=head1 METHODS

=cut

package Bio::EnsEMBL::ChainedAssemblyMapper;
$Bio::EnsEMBL::ChainedAssemblyMapper::VERSION = '114.0.0';
use strict;
use warnings;
use integer; #use proper arithmetic bitshifts

use Bio::EnsEMBL::Mapper;
use Bio::EnsEMBL::Mapper::RangeRegistry;
use Bio::EnsEMBL::Utils::Exception qw(throw);
use Scalar::Util qw(weaken);
use Bio::EnsEMBL::Utils::Scalar qw( check_ref);

my $FIRST = 'first';
my $MIDDLE = 'middle';
my $LAST  = 'last';

#2^20 = approx 10^6
my $CHUNKFACTOR = 20;

# max size of the pair cache in the mappers
my $DEFAULT_MAX_PAIR_COUNT = 6000;

=head2 new

  Arg [1]    : Bio::EnsEMBL::DBSQL::AssemblyMapperAdaptor
  Arg [2]    : Bio::EnsEMBL::CoordSystem $src_cs
  Arg [3]    : Bio::EnsEMBL::CoordSystem $int_cs
  Arg [4]    : Bio::EnsEMBL::CoordSystem $dst_cs
  Example    : Should use AssemblyMapperAdaptor->fetch_by_CoordSystems
  Description: Creates a new AssemblyMapper
  Returntype : Bio::EnsEMBL::DBSQL::AssemblyMapperAdaptor
  Exceptions : thrown if wrong number of coord_systems are provided
  Caller     : AssemblyMapperAdaptor
  Status     : Stable

=cut

sub new {
  my ($caller,$adaptor,@coord_systems) = @_;

  my $class = ref($caller) || $caller;

  my $self = {};
  bless $self, $class;

  $self->adaptor($adaptor);

  if(@coord_systems != 3) {
    throw('ChainedMapper can only map between 3 coordinate systems. ' .
          scalar(@coord_systems) . ' were provided');
  }

  $adaptor->cache_seq_ids_with_mult_assemblys();