view release on metacpan or  search on metacpan

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

=cut

=head1 NAME

Bio::EnsEMBL::AssemblyMapper - 
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 $ctg_cs = $cs_adaptor->fetch_by_name('contig');

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

    # Map to contig coordinate system from chromosomal.
    @ctg_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( 'AL30421.1.200.92341', 100, 10000, -1,
      $ctg_cs );

    # List contig names for a region of chromsome.
    @ctg_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( 'AL30421.1.200.92341', 1, 1000, -1,
      $ctg_cs );

=head1 DESCRIPTION

The AssemblyMapper is a database aware mapper which faciliates
conversion of coordinates between any two coordinate systems with an
relationship explicitly defined in the assembly table.  In the future
it may be possible to perform multiple step (implicit) mapping between
coordinate systems.

It is implemented using the Bio::EnsEMBL::Mapper object, which is a
generic mapper object between disjoint coordinate systems.

=head1 METHODS

=cut


package Bio::EnsEMBL::AssemblyMapper;
$Bio::EnsEMBL::AssemblyMapper::VERSION = '114.0.0';
use strict;
use warnings;

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

my $ASSEMBLED = 'assembled';
my $COMPONENT = 'component';

my $DEFAULT_MAX_PAIR_COUNT = 1000;


=head2 new

  Arg [1]    : Bio::EnsEMBL::DBSQL::AssemblyMapperAdaptor
  Arg [2]    : Bio::EnsEMBL::CoordSystem $asm_cs
  Arg [3]    : Bio::EnsEMBL::CoordSystem $cmp_cs
  Example    : Should use AssemblyMapperAdaptor->fetch_by_CoordSystems()
  Description: Creates a new AssemblyMapper
  Returntype : Bio::EnsEMBL::DBSQL::AssemblyMapperAdaptor
  Exceptions : Throws if multiple coord_systems are provided
  Caller     : AssemblyMapperAdaptor
  Status     : Stable

=cut

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

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

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

  $self->adaptor($adaptor);

  $adaptor->cache_seq_ids_with_mult_assemblys();

  if ( @coord_systems != 2 ) {
    throw(   'Can only map between two coordinate systems. '
           . scalar(@coord_systems)
           . ' were provided' );
  }

  # Set the component and assembled coordinate systems
  $self->{'asm_cs'} = $coord_systems[0];
  $self->{'cmp_cs'} = $coord_systems[1];

  # We load the mapper calling the 'ASSEMBLED' the 'from' coord system
  # and the 'COMPONENT' the 'to' coord system.

  $self->{'mapper'} = Bio::EnsEMBL::Mapper->new( $ASSEMBLED, $COMPONENT,
                                 $coord_systems[0], $coord_systems[1] );

  $self->{'max_pair_count'} = $DEFAULT_MAX_PAIR_COUNT;

  return $self;
} ## end sub new

=head2 max_pair_count

  Arg [1]    : (optional) int $max_pair_count
  Example    : $mapper->max_pair_count(100000)
  Description: Getter/Setter for the number of mapping pairs allowed
               in the internal cache.  This can be used to override

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

sub mapper {
  my ($self) = @_;

  return $self->{'mapper'};
}

=head2 assembled_CoordSystem

  Arg [1]    : None
  Example    : $cs = $asm_mapper->assembled_CoordSystem();
  Description: Retrieves the assembled CoordSystem from this
               assembly mapper.
  Return type: Bio::EnsEMBL::CoordSystem
  Exceptions : None
  Caller     : Internal, AssemblyMapperAdaptor
  Status     : Stable

=cut

sub assembled_CoordSystem {
  my ($self) = @_;

  return $self->{'asm_cs'};
}

=head2 component_CoordSystem

  Arg [1]    : None
  Example    : $cs = $asm_mapper->component_CoordSystem();
  Description: Retrieves the component CoordSystem from this
               assembly mapper.
  Return type: Bio::EnsEMBL::CoordSystem
  Exceptions : None
  Caller     : Internal, AssemblyMapperAdaptor
  Status     : Stable

=cut

sub component_CoordSystem {
  my ($self) = @_;

  return $self->{'cmp_cs'};
}

=head2 adaptor

  Arg [1]    : Bio::EnsEMBL::DBSQL::AssemblyMapperAdaptor $adaptor
  Description: Getter/set terfor this object's database adaptor.
  Returntype : Bio::EnsEMBL::DBSQL::AssemblyMapperAdaptor
  Exceptions : None
  Caller     : General
  Status     : Stable

=cut

sub adaptor {
  my ( $self, $value ) = @_;

  if ( defined($value) ) {
    weaken($self->{'adaptor'} = $value);
  }

  return $self->{'adaptor'};
}

1;