Toolbar

Summary

Bio::EnsEMBL::Analysis::Runnable::Spliced_elsewhere;

Package variables

Included modules

Inherit

Bio::EnsEMBL::Analysis::Runnable::Pseudogene

Synopsis

  my $runnable = Bio::EnsEMBL::Analysis::Runnable::Spliced_elsewhere ->new
  (
   '-genes' => \@genes_array_ref,
   '-analysis' => $analysis_object,
  );
    $runnable->run;
    $output = $runnable->output

Description

Runnable for Bio::EnsEMBL::Analysis::RunnableDB::Spliced_elsewhere
Does the blast analysis and returns the results to the runnable_db for
parsing.
Uses Bio::EnsEMBL::Analysis::Config::Pseudogene for config

Methods

Methods description

  Arg [1]    : scalar
  Description: get/set path to blast database
  Returntype : scalar
  Exceptions : none
  Caller     : general

  Args       : various
  Description: Runnable constructor
  Returntype : Bio::EnsEMBL::Analysis::Runnable::Spliced_elsewhere;
  Caller     : general

  Arg [1]    : hasref
  Description: overrides output array
  Returntype : array
  Exceptions : none
  Caller     : general

Methods code

sub db {

  my ($self, $db) = @_;
  if ($db) {
    $self->find_file($db);
    $self->{'_db'} = $db;
  }
  return $self->{'_db'};

}

sub new {

  my ($class,@args) = @_;
  my $self = $class->SUPER::new(@args);

  $self->{'_genes'} = [];	#array of genescripts to test;  

  my($genes,$PS_MULTI_EXON_DIR) =
    rearrange([qw( GENES
		   PS_MULTI_EXON_DIR
 )], @args);
 
  $self->genes($genes);
  $self->PS_MULTI_EXON_DIR($PS_MULTI_EXON_DIR);

  # Path to blast database
  $self->db($self->PS_MULTI_EXON_DIR."/all_multi_exon_genes.fasta");
  return $self;

}

sub output {

  my ($self, $hash_ref) = @_;
  if ($hash_ref) {
    push @{$self->{'_output'}},$hash_ref;
  }
  return $self->{'_output'};

}

sub run {

  my ($self)=@_;
  my @genes = @{$self->genes};
 GENE: foreach my $gene(@genes){
    $self->run_blast($gene);
    next GENE;
  }
  return 1;

}

sub run_blast {

  my ($self,$gene)=@_;
  my $bplitewrapper = Bio::EnsEMBL::Analysis::Tools::BPliteWrapper-> new
    (
     -query_type => 'dna',
     -database_type => 'dna',
    );
  my %output_hash;
  foreach my $trans (@{$gene->get_all_Transcripts}){ 
    next unless ($trans->translateable_seq);
    my $query = $trans->feature_Slice->expand(1000,1000)->get_repeatmasked_seq;
    my $test = $trans->feature_Slice->get_repeatmasked_seq->seq;
  #  print "BEFORE  Ns have gone " . length($test) . "\n"; 
    $test =~ s/N//g;
#    print "AFTER Ns have gone " . length($test) . "\n";
    unless ( length($test) > 5 ) {
      print STDOUT "ignoring " . $trans->display_id. " all Ns\n ";
      $output_hash{$trans->dbID}= 'REPEATS';
      $self->output(\%output_hash);
    return 1;
    }
    print  $trans->stable_id if ( $trans->stable_id ) ; 
    # need to mask out other coding sequences here
    my $blast =  Bio::EnsEMBL::Analysis::Runnable::Blast->new 
      ('-query'     => $query,
       '-program'   => 'blastn',
       '-database'  => $self->db,
       '-threshold' => 1e-50,
       '-parser'    => $bplitewrapper,
       '-options'   => 'V=10',
       '-analysis'  => $self->analysis,
      );
    eval {
      $blast->run();
    };
    if ($@) {
      $self->throw("Problem with Blast $@\n  ");
    } else {
      if ( $blast->output ) {
	$output_hash{$trans->dbID}= $blast->output;
      } else {
	$output_hash{$trans->dbID}= 'NONE';
      }
    }
  }
  $self->output(\%output_hash);
  return 1;
}



########################################################
# Containers

}

General documentation

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