Toolbar

Summary

Bio::EnsEMBL::Analysis::RunnableDB::Exonerate2Genes;

Package variables

Included modules

Inherit

Bio::EnsEMBL::Analysis::RunnableDB

Synopsis

my $clone =
Bio::EnsEMBL::Analysis::RunnableDB::ExonerateConeEnds->new(
-db => $refdb,
-analysis => $analysis_obj,
-database => $EST_GENOMIC,
-query_seqs => \@sequences,
);
$clone->fetch_input();
$clone->run();
$clone->write_output(); #writes to DB

Description

This object maps clone sequences to a genome,
and writing the results as Features.

Methods

Methods description

None available.

Methods code

sub DNADB {

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

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

  if ( exists( $self->{'_CONFIG_DNADB'} ) ) {
    return $self->{'_CONFIG_DNADB'};
  } else {
    return undef;
  }

}

sub GENOMICSEQS {

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

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

  if ( exists( $self->{'_CONFIG_GENOMICSEQS'} ) ) {
    return $self->{'_CONFIG_GENOMICSEQS'};
  } else {
    return undef;
  }

}

sub IIDREGEXP {

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

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

  if ( exists( $self->{'_CONFIG_IIDREGEXP'} ) ) {
    return $self->{'_CONFIG_IIDREGEXP'};
  } else {
    return undef;
  }

}

sub OPTIONS {

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

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

  if ( exists( $self->{'_CONFIG_OPTIONS'} ) ) {
    return $self->{'_CONFIG_OPTIONS'};
  } else {
    return undef;
  }
}

###############################################
###     end of config
###############################################

1;

}

sub OUTDB {

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

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

  if ( exists( $self->{'_CONFIG_OUTDB'} ) ) {
    return $self->{'_CONFIG_OUTDB'};
  } else {
    return undef;
  }

}

sub QUERYSEQS {

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

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

  if ( exists( $self->{'_CONFIG_QUERYSEQS'} ) ) {
    return $self->{'_CONFIG_QUERYSEQS'};
  } else {
    return undef;
  }

}

sub create_output_db {

  my ($self) = @_;

  my $outdb;
  my $dnadb;

  if ( $self->OUTDB) {
    $outdb = new Bio::EnsEMBL::DBSQL::DBAdaptor(%{ $self->OUTDB }); 
    if ($self->DNADB) {
      $outdb->dnadb(new Bio::EnsEMBL::DBSQL::DBAdaptor(%{$self->DNADB})); 
    }
  } else {
    $outdb = $self->db;
  }

  return $outdb;
}


#############################################################
# Declare and set up config variables
#############################################################

}

sub fetch_input {

  my ($self) = @_;

  my $logic = $self->analysis->logic_name;

  ##########################################
  # set up the target (genome)
  ##########################################

  my $target = $self->GENOMICSEQS;
  my $query =  $self->QUERYSEQS;

  if ( -e $target ){
    if(-d $target ) {
      warn ("Target $target is a directory of files\n");
    }elsif (-s $target){
      warn ("Target $target is a whole-genome file\n");
    }else{
      throw("'$target' isn't a file or a directory?");
    }
  } else {
    throw("'$target' could not be found");
  }
  if ( -e $query){
    if(-d $query ) {
      warn ("Query $query is a directory of files\n");
    }elsif (-s $target){
      warn ("Query $query is a whole-genome file\n");
    }else{
      throw("'$query' isn't a file or a directory?");
    }
  } else {
    throw("'$query' could not be found");
  }

  ##########################################
  # set up the query (dna clone seq)
  ##########################################

  my $iid_regexp = $self->IIDREGEXP;
  my ($chunk_number, $chunk_total ) = $self->input_id =~ /$iid_regexp/;
  if(!$chunk_number || !$chunk_total){
    throw "I can't make sense of your input id  using the IIDREGEXP in the config!\n";
  }

  ##########################################
  # setup the runnable
  ##########################################

  my %parameters = %{ $self->parameters_hash };
  
  if (not exists( $parameters{-options} )
      and defined $self->OPTIONS) {
    $parameters{-options} = $self->OPTIONS;
  }
  
  #print STDERR "PROGRAM FILE: ".$self->analysis->program_file."\n";
  
  my $runnable = Bio::EnsEMBL::Analysis::Runnable::ExonerateCloneEnds->new(
    -program            => $self->analysis->program_file,
    -analysis           => $self->analysis,
    -target_file        => $target,
    -query_type         => 'dna',
    -query_file         => $query,
    -query_chunk_number => $chunk_number,
    -query_chunk_total  => $chunk_total,
    %parameters,
  );
  
  $self->runnable($runnable);

}

############################################################

}

sub new {

  my ( $class, @args ) = @_;
  my $self = $class->SUPER::new(@args);
  
  #Because we dont know whether the sort of adapter (compara, hive, core)
  #we'll be passed, it's better to just remake the core dbadaptor and
  #keep it on ourselves as the dna db - this has to be the intent of the
  #dataadaptor input through the rulemanager / beekeeper / whatever

  $self->read_and_check_config($CLONE_CONFIG);
  return $self;

}

sub process_features {

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

  my $db = $self->create_output_db;
  my $slice_adaptor = $db->get_SliceAdaptor;

  my %genome_slices;

  foreach my $f (@$flist) {

    $f->analysis( $self->analysis );

    # get the slice based on the seqname stamped on in the runnable
    my $slice_id = $f->seqname;

    if ( not exists $genome_slices{$slice_id} ) {
      # assumes genome seqs were named in the Ensembl API Slice naming
      # convention, i.e. coord_syst:version:seq_reg_id:start:end:strand
      $genome_slices{$slice_id} = $slice_adaptor->fetch_by_name($slice_id);
    }
    my $slice = $genome_slices{$slice_id};

    $f->slice($slice);  
  }

}

sub read_and_check_config {

  my $self = shift;

  $self->SUPER::read_and_check_config($CLONE_CONFIG);

  ##########
  # CHECKS
  ##########
  my $logic = $self->analysis->logic_name;

  # check that compulsory options have values
  foreach my $config_var (
    qw(
      QUERYSEQS
      GENOMICSEQS
    )
  ){
    if ( not defined $self->$config_var ){
      throw("You must define $config_var in config for logic '$logic'");
    }
  }

  # output db does not have to be defined, but if it is, it should be a hash
  if ($self->OUTDB && ref( $self->OUTDB ) ne "HASH") {
    throw("OUTDB in config for '$logic' must be a hash ref of db connection pars.");
  }
  
  if ( $self->DNADB and ref( $self->DNADB ) ne "HASH" ) {
    throw("DNADB in config for '$logic' must be a hash ref of db connection pars.");
  }

}

sub run {

  my ($self) = @_;

  throw("Can't run - no runnable objects") unless ( $self->runnable );

  my ($runnable) = @{$self->runnable};
  
  $runnable->run;

  my $features = $runnable->output;
  $self->process_features($runnable->output);

  $self->output($features);
}

############################################################

}

sub write_output {

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

  my $outdb = $self->create_output_db;
  my $fa = $outdb->get_DnaAlignFeatureAdaptor;

  foreach my $f (@{$self->output}){  
    eval{ 
      $fa->store($f);
    };
    if ($@) {
      $self->throw("Unable to store clone feature!\n $@");
    }
  }
}

############################################################

}

General documentation

Post general queries to ensembl-dev@ebi.ac.uk