BasicMapper.pm

package XrefMapper::BasicMapper;

use strict;
use Cwd;
use DBI;
use File::Basename;
use IPC::Open3;
use Bio::EnsEMBL::DBSQL::DBAdaptor;
use Bio::EnsEMBL::Translation;
use XrefMapper::db;
use vars '@ISA';


=head1 NAME

XrefMapper::BasicMapper

=head1 DESCIPTION

This is the basic mapper routine. It will create the necessary fasta files for
both the xref and ensembl sequences. These will then be matched using exonerate
and the results written to another file. By creating a <species>.pm file and 
inheriting from this base class different matching routines, parameters, data 
sets etc can be set.

=head1 CONTACT

Post questions to the EnsEMBL development list ensembl-dev@ebi.ac.uk


=cut

# Hashes to hold method-specific thresholds
my %method_query_threshold;
my %method_target_threshold;

# Various useful variables.
my %translation_to_transcript;
my %transcript_to_translation;
my %genes_to_transcripts;
my %xref_to_source;
my %object_xref_mappings;
my %object_xref_identities;
my %xref_descriptions;
my %xref_accessions;
my %xref_labels;
my %source_to_external_db;
my %xrefs_written;
my %object_xrefs_written;
my %failed_xref_mappings;
my %updated_source;

=head2 new

  Description: Constructor for BasicMapper.
  Returntype : BasicMapper
  Exceptions : none
  Caller     : general

=cut

sub new{
  my($class, @args) = @_;

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

  return $self;
}


=head2 dump_seqs

  Arg[1]: xref object which holds info needed for the dump of xref

  Description: Dumps out the files for the mapping. Xref object should hold
              the value of the databases and source to be used.
  Returntype : none
  Exceptions : will die if species not known or an error occurs while
             : trying to write to files. 
  Caller     : general
 
=cut
 


sub dump_seqs{

  my ($self, $location) = @_;

  $self->dump_xref();
  $self->dump_ensembl($location);

}



=head2 build_list_and_map

  Arg[1]: xref object which holds info on method and files.

  Description: runs the mapping of the list of files with species methods
  Returntype : none
  Exceptions : none
  Caller     : general

=cut

sub build_list_and_map {

  my ($self) = @_;

  my @list=();

  my $i = 0;
  foreach my $method (@{$self->method()}){
    my @dna=();
    push @dna, $method;
    push @dna, $self->xref->dir."/xref_".$i."_dna.fasta";
    push @dna, $self->core->dna_file();
    push @list, \@dna;
    my @pep=();
    push @pep, $method;
    push @pep, $self->xref->dir."/xref_".$i."_peptide.fasta";
    push @pep, $self->core->protein_file();
    push @list, \@pep;
    $i++;
  }

  $self->run_mapping(\@list);

}


=head2 get_species_id_from_species_name

  Arg[1]: species name

  Description: get the species_id from the database for the named database.
  Example    : my $id = get_species_id_from_species_name('homo_sapiens');
  Returntype : int (species_id)
  Exceptions : will die if species does not exist in given xref database.
  Caller     : general

=cut

sub get_species_id_from_species_name{
  my ($self,$species) = @_;


  my $sql = "select species_id from species where name = '".$species."'";
  my $sth = $self->dbc->prepare($sql);
  $sth->execute();
  my @row = $sth->fetchrow_array();
  my $species_id;
  if (@row) {
    $species_id = $row[0];
  } else {
    print STDERR "Couldn't get ID for species ".$species."\n";
    print STDERR "It must be one of :-\n";
    $sql = "select name from species";
    $sth = $self->dbc->prepare($sql);
    $sth->execute();
    while(my @row = $sth->fetchrow_array()){
      print STDERR $row[0]."\n";
    }
    die("Please try again :-)\n");
  }
  $sth->finish();

  return $species_id;
}


=head2 get_set_lists

  Description: specifies the list of databases and source to be used in the
             : generation of one or more data sets.
  Returntype : list of lists
  Example    : my @lists =@{$self->get_set_lists()};
  Exceptions : none
  Caller     : dump_xref

=cut

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

  return [["ExonerateGappedBest1", ["*","*"]]];

}

=head2 get_source_id_from_source_name

  Arg[1]: source name

  Description: get the source_id from the database for the named source.
  Example    : my $id = get_source_id_from_source_name('RefSeq');
  Returntype : int (source_id)
  Exceptions : will die if source does not exist in given xref database.