Inheritance diagram for Bio::EnsEMBL::IdMapping::GeneScoreBuilder:

Public Member Functions
public	score_genes ()

public	scores_from_transcript_scores ()

public	build_scores ()

public	flag_matrix_from_transcript_scores ()

public	score_matrix_from_flag_matrix ()

public	complex_gene_gene_score ()

public	simple_gene_gene_score ()

public	simple_gene_rescore ()

public	biotype_gene_rescore ()

public	location_gene_rescore ()

Public Member Functions inherited from Bio::EnsEMBL::IdMapping::ScoreBuilder
public Bio::EnsEMBL::IdMapping::ScoredMappingMatrix	create_shrinked_matrix ()

public void	internal_id_rescore ()

public void	log_matrix_stats ()

Public Member Functions inherited from Bio::EnsEMBL::IdMapping::BaseObject
public $this	new ()

public Filehandle	get_filehandle ()

public Boolean	file_exists ()

public SQLStatement	fetch_value_from_db ()

public Int	dump_table_to_file ()

public Int	upload_file_into_table ()

public Bio::EnsEMBL::Utils::Logger	logger ()

public Bio::EnsEMBL::Utils::ConfParser	conf ()

public Bio::EnsEMBL::IdMapping::Cache	cache ()

Detailed Description

Synopsis

Description

Combines ExonScoreBuilder, ExonDirectMapper and ExonerateRunner from
Java application.

Definition at line 18 of file GeneScoreBuilder.pm.

Member Function Documentation

◆ biotype_gene_rescore()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::biotype_gene_rescore ( )

Undocumented method

Code:

click to view

◆ build_scores()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::build_scores ( )

Undocumented method

Code:

click to view

◆ complex_gene_gene_score()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::complex_gene_gene_score ( )

Undocumented method

Code:

click to view

sub complex_gene_gene_score {
  my $self = shift;
  my $source_gene = shift;
  my $target_gene = shift;
  my $transcript_matrix = shift;
 
  my $source_gene_score = 0;
  my $target_gene_score = 0;
  my $source_gene_accum_length = 0; # sum of all transcript lengths
  my $target_gene_accum_length = 0; # sum of all transcript lengths
 
  # We are only interested in scoring with transcripts that are in the target
  # gene. The scored mapping matrix may contain scores for transcripts that
  # aren't in this gene so create a hash of the target genes's transcripts
  my %target_transcripts = map { $_->id => 1 }
    @{ $target_gene->get_all_Transcripts };
    
  # loop over source transcripts
  foreach my $source_transcript (@{ $source_gene->get_all_Transcripts }) {
 
    # now loop over target transcripts and find the highest scoring target
    # transcript belonging to the target gene
    my $max_source_score = -1;
    
    foreach my $target_transcript_id (@{ $transcript_matrix->get_targets_for_source($source_transcript->id) }) {
 
      next unless ($target_transcripts{$target_transcript_id});
 
      my $score = $transcript_matrix->get_score(
        $source_transcript->id, $target_transcript_id);
      $max_source_score = $score if ($score > $max_source_score);
    }
 
    if ($max_source_score > 0) {
      $source_gene_score += ($max_source_score * $source_transcript->length);
    }
 
    $source_gene_accum_length += $source_transcript->length;
  }
    
  # now do the same for target genes
  my %source_transcripts = map { $_->id => 1 }
    @{ $source_gene->get_all_Transcripts };
    
  # loop over target transcripts
  foreach my $target_transcript (@{ $target_gene->get_all_Transcripts }) {
 
    # now loop over source transcripts and find the highest scoring source
    # transcript belonging to the source gene
    my $max_target_score = -1;
    
    foreach my $source_transcript_id (@{ $transcript_matrix->get_sources_for_target($target_transcript->id) }) {
 
      next unless ($source_transcripts{$source_transcript_id});
 
      my $score = $transcript_matrix->get_score(
        $source_transcript_id, $target_transcript->id);
      $max_target_score = $score if ($score > $max_target_score);
    }
 
    if ($max_target_score > 0) {
      $target_gene_score += ($max_target_score * $target_transcript->length);
    }
 
    $target_gene_accum_length += $target_transcript->length;
  }
 
  # calculate overall score for this gene
  my $gene_score = 0;
 
  if (($source_gene_accum_length + $target_gene_accum_length) > 0) {
 
    $gene_score = ($source_gene_score + $target_gene_score) /
                  ($source_gene_accum_length + $target_gene_accum_length);
 
  } else {
  
    $self->logger->warning("Combined transcript length of source (".$source_gene->id.") and target (".$target_gene->id.") gene is zero!\n", 1);
  
  }
 
  if ($gene_score > 1) {
    $self->logger->warning("Illegal gene score: $gene_score (".
      join("|", $source_gene_score, $target_gene_score,
           $source_gene_accum_length, $target_gene_accum_length).")\n", 1);
  }
 
  return $gene_score;
}

◆ flag_matrix_from_transcript_scores()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::flag_matrix_from_transcript_scores ( )

Undocumented method

Code:

click to view

◆ location_gene_rescore()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::location_gene_rescore ( )

Undocumented method

Code:

click to view

◆ score_genes()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::score_genes ( )

Undocumented method

Code:

click to view

◆ score_matrix_from_flag_matrix()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::score_matrix_from_flag_matrix ( )

Undocumented method

Code:

click to view

sub score_matrix_from_flag_matrix {
  my $self              = shift;
  my $flag_matrix       = shift;
  my $transcript_matrix = shift;
  my $simple_scoring    = shift;
 
  unless ( $flag_matrix and
     $flag_matrix->isa('Bio::EnsEMBL::IdMapping::ScoredMappingMatrix') )
  {
    throw('Need a gene Bio::EnsEMBL::IdMapping::ScoredMappingMatrix.');
  }
 
  unless (     $transcript_matrix
           and
           $transcript_matrix->isa(
                         'Bio::EnsEMBL::IdMapping::ScoredMappingMatrix')
    )
  {
    throw(
       'Need a transcript Bio::EnsEMBL::IdMapping::ScoredMappingMatrix.'
    );
  }
 
  # create a new scoring matrix which will replace the flag matrix
  my $matrix =
    Bio::EnsEMBL::IdMapping::ScoredMappingMatrix->new(
                           -DUMP_PATH  => $flag_matrix->dump_path,
                           -CACHE_FILE => $flag_matrix->cache_file_name,
    );
 
  # initialise progress logger
  my $i;
  my $num_entries = $flag_matrix->get_entry_count;
  my $progress_id = $self->logger->init_progress( $num_entries, 100 );
 
  $self->logger->info( "Creating score matrix from flag matrix...\n",
                       1 );
 
  my $gene_score_threshold =
    $self->conf()->param('gene_score_threshold') || 0;
 
  # loop over flag matrix and do proper scoring for each entry
  foreach my $entry ( @{ $flag_matrix->get_all_Entries } ) {
 
    $self->logger->log_progress( $progress_id, ++$i, 1 );
 
    my $score = 0;
    my $source_gene =
      $self->cache->get_by_key( 'genes_by_id', 'source',
                                $entry->source );
    my $target_gene =
      $self->cache->get_by_key( 'genes_by_id', 'target',
                                $entry->target );
 
    if ($simple_scoring) {
 
      # simple scoring (used for rescoring purposes)
      $score =
        $self->simple_gene_gene_score( $source_gene, $target_gene,
                                       $transcript_matrix );
 
    }
    else {
 
      # full scoring
      $score =
        $self->complex_gene_gene_score( $source_gene, $target_gene,
                                        $transcript_matrix );
    }
 
    if ( $score > $gene_score_threshold ) {
      $matrix->add_score( $entry->source, $entry->target, $score );
    }
  } ## end foreach my $entry ( @{ $flag_matrix...})
 
  $self->logger->info("\n");
 
  return $matrix;
} ## end sub score_matrix_from_flag_matrix

◆ scores_from_transcript_scores()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::scores_from_transcript_scores ( )

Undocumented method

Code:

click to view

◆ simple_gene_gene_score()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::simple_gene_gene_score ( )

Undocumented method

Code:

click to view

◆ simple_gene_rescore()

public Bio::EnsEMBL::IdMapping::GeneScoreBuilder::simple_gene_rescore ( )

Undocumented method

Code:

click to view

The documentation for this class was generated from the following file:

ensembl/modules/Bio/EnsEMBL/IdMapping/GeneScoreBuilder.pm

Public Member Functions

Detailed Description

Synopsis

Description

Member Function Documentation

◆ biotype_gene_rescore()

◆ build_scores()

◆ complex_gene_gene_score()

◆ flag_matrix_from_transcript_scores()

◆ location_gene_rescore()

◆ score_genes()

◆ score_matrix_from_flag_matrix()

◆ scores_from_transcript_scores()

◆ simple_gene_gene_score()

◆ simple_gene_rescore()