Public Member Functions
public	new ()

public	get_official_name ()

public	run ()

public	get_official_domain_name ()

public	set_transcript_display_xrefs ()

public	get_dbentrie_sth ()

public	get_dbentrie_with_desc_sth ()

public	get_best_identity_sth ()

public	get_set_transcript_display_xref_sth ()

public	get_set_gene_display_xref_sth ()

public	get_ins_xref_sth ()

public	get_ins_dep_ix_sth ()

public	get_ins_object_xref_sth ()

public	find_max_ids ()

public	get_synonyms ()

public	get_display_label_data ()

public	get_other_name_hash ()

public	find_from_other_sources ()

public	get_delete_odn_sth ()

public	set_the_best_odns ()

public	get_lrg_find_sth ()

public	get_lrg_set_status_sth ()

public	get_lrg_to_hgnc_sth ()

public	find_lrg_hgnc ()

public	get_new_dbname_sources ()

public	delete_old_data ()

public	reset_display_xrefs ()

Detailed Description

Definition at line 6 of file OfficialNaming.pm.

Member Function Documentation

◆ delete_old_data()

public XrefMapper::OfficialNaming::delete_old_data ( )

Undocumented method

Code:

click to view

◆ find_from_other_sources()

public XrefMapper::OfficialNaming::find_from_other_sources ( )

Undocumented method

Code:

click to view

◆ find_lrg_hgnc()

public XrefMapper::OfficialNaming::find_lrg_hgnc ( )

Undocumented method

Code:

click to view

◆ find_max_ids()

public XrefMapper::OfficialNaming::find_max_ids ( )

Undocumented method

Code:

click to view

◆ get_best_identity_sth()

public XrefMapper::OfficialNaming::get_best_identity_sth ( )

Undocumented method

Code:

click to view

◆ get_dbentrie_sth()

public XrefMapper::OfficialNaming::get_dbentrie_sth ( )

Undocumented method

Code:

click to view

◆ get_dbentrie_with_desc_sth()

public XrefMapper::OfficialNaming::get_dbentrie_with_desc_sth ( )

Undocumented method

Code:

click to view

◆ get_delete_odn_sth()

public XrefMapper::OfficialNaming::get_delete_odn_sth ( )

Undocumented method

Code:

click to view

◆ get_display_label_data()

public XrefMapper::OfficialNaming::get_display_label_data ( )

Undocumented method

Code:

click to view

◆ get_ins_dep_ix_sth()

public XrefMapper::OfficialNaming::get_ins_dep_ix_sth ( )

Undocumented method

Code:

click to view

◆ get_ins_object_xref_sth()

public XrefMapper::OfficialNaming::get_ins_object_xref_sth ( )

Undocumented method

Code:

click to view

◆ get_ins_xref_sth()

public XrefMapper::OfficialNaming::get_ins_xref_sth ( )

Undocumented method

Code:

click to view

◆ get_lrg_find_sth()

public XrefMapper::OfficialNaming::get_lrg_find_sth ( )

Undocumented method

Code:

click to view

◆ get_lrg_set_status_sth()

public XrefMapper::OfficialNaming::get_lrg_set_status_sth ( )

Undocumented method

Code:

click to view

◆ get_lrg_to_hgnc_sth()

public XrefMapper::OfficialNaming::get_lrg_to_hgnc_sth ( )

Undocumented method

Code:

click to view

◆ get_new_dbname_sources()

public XrefMapper::OfficialNaming::get_new_dbname_sources ( )

Undocumented method

Code:

click to view

◆ get_official_domain_name()

public XrefMapper::OfficialNaming::get_official_domain_name ( )

Undocumented method

Code:

click to view

sub get_official_domain_name{
  my ($self, $arg_ref) = @_;
 
  my $gene_id              = $arg_ref->{gene_id};
  my $gene_id_to_stable_id = $arg_ref->{gene_id_to_stable_id};
  my $gene_to_transcripts  = $arg_ref->{gene_to_tran};
  my $official_name_used   = $arg_ref->{official_name_used};
  my $dbi                  = $arg_ref->{dbi};
 
 
  my $dbname = $self->get_official_name();
  my $gene_symbol = undef;
  my $gene_symbol_xref_id = undef;
 
 
  my $dbentrie_sth = $self->get_dbentrie_sth($dbi);
 
  my %ODN=();
  my %xref_id_to_display;
 
  $dbentrie_sth->execute($dbname, $gene_id, "Gene");
  my ($display, $xref_id, $object_xref_id, $level); 
  $dbentrie_sth->bind_columns(\$display, \$xref_id, \$object_xref_id, \$level);
  my $best_level=999;
  
  my $count = 0;
  my @list=();
  my @list_ox=();
 
  while($dbentrie_sth->fetch){
 
    push @list, $xref_id;
    push @list_ox, $object_xref_id;
    $count++;
    $xref_id_to_display{$xref_id} = $display;
    if($level < $best_level){
      %ODN = ();
      $ODN{$xref_id} = 1;
      $best_level = $level;
    }
    elsif($level == $best_level){
      $ODN{$xref_id} = 1;
    }
  }
 
  if(($count > 1) and (scalar(keys %ODN) == 1)){ # found one that is "best" so set it and remove others
    print "For gene ".$gene_id_to_stable_id->{$gene_id}." we have mutiple ".$dbname."'s\n";
    ($gene_symbol, $gene_symbol_xref_id) = $self->set_the_best_odns(\%ODN, \@list, \@list_ox, \%xref_id_to_display, $dbi);
    if(defined($gene_symbol)){
      return $gene_symbol, $gene_symbol_xref_id;
    }
  }
 
  if(scalar(keys %ODN) == 1){  # one hgnc to this gene - perfect case :-)
      return $xref_id_to_display{(keys %ODN)[0]}, (keys %ODN)[0];
  }
  if(scalar(keys %ODN) > 1){ 
    
    #if we have  more than 1 xref, fail xrefs with worse % identity if we can (query or target identity whichever is greater)
    my $identity_sth = $self->get_best_identity_sth($dbi);     
    $identity_sth->execute($dbname, $gene_id, "Gene");
    my ($xref_id, $best_identity); 
    $identity_sth->bind_columns(\$xref_id, \$best_identity);
    my $temp_best_identity = 0;
    my %best_ids = ();
 
    while($identity_sth->fetch){
     
    if($best_identity > $temp_best_identity){
         %best_ids = ();
         $best_ids{$xref_id} = 1;
         $temp_best_identity = $best_identity;
    }
    elsif($best_identity == $temp_best_identity){
         $best_ids{$xref_id} = 1;
    } 
    else {
         last;
    }
    }
 
    my %best_list;
    foreach my $xref_id (keys %ODN){
      $best_list{$xref_id_to_display{$xref_id}} = 1;
    }
 
    # check if we were able to reduce the number of xrefs based on % identity 
    if ( scalar(keys %best_ids) > 0 && scalar(keys %best_ids) < scalar(keys %ODN) ) {
      %ODN = %best_ids;
      print "For gene ".$gene_id_to_stable_id->{$gene_id}." we have mutiple ".$dbname."'s\n";
      #set statuses for xrefs with worse % identity to MULTI_DELETE
      ($gene_symbol, $gene_symbol_xref_id) = $self->set_the_best_odns(\%ODN, \@list, \@list_ox, \%xref_id_to_display, $dbi);
      if( defined($gene_symbol) && scalar(keys %ODN == 1) ){
          return $gene_symbol, $gene_symbol_xref_id;
      } 
    }
    
    # take the name which hasn't been already assigned to another gene, if possible
    
    my $xref_not_used;
    foreach my $x (keys %ODN){
      if (!defined($official_name_used->{$x}) ) {
          $xref_not_used = $x;
      }
    }
    if ($xref_not_used) {
      foreach my $x (keys %ODN){
          print "\t".$xref_id_to_display{$x};
          if ($x == $xref_not_used) {
          print "    chosen\n";
          $gene_symbol =  $xref_id_to_display{$x};
          $gene_symbol_xref_id =  $x;         
          } else {
          print "  (left as $dbname reference but not gene symbol)\n";
          }
      }
 
    } else {
 
      my $i=0;
      foreach my $x (keys %ODN){
          print "\t".$xref_id_to_display{$x};
          if(!$i){
          print "  (chosen as first)\n";
          $gene_symbol =  $xref_id_to_display{$x};
          $gene_symbol_xref_id =  $x;
          }
          else{
          print "  (left as $dbname reference but not gene symbol)\n";
          }
          $i++;
      }
 
 
    }
  }
  return ($gene_symbol, $gene_symbol_xref_id);
}

◆ get_official_name()

public XrefMapper::OfficialNaming::get_official_name ( )

Undocumented method

Code:

click to view

◆ get_other_name_hash()

public XrefMapper::OfficialNaming::get_other_name_hash ( )

Undocumented method

Code:

click to view

◆ get_set_gene_display_xref_sth()

public XrefMapper::OfficialNaming::get_set_gene_display_xref_sth ( )

Undocumented method

Code:

click to view

◆ get_set_transcript_display_xref_sth()

public XrefMapper::OfficialNaming::get_set_transcript_display_xref_sth ( )

Undocumented method

Code:

click to view

◆ get_synonyms()

public XrefMapper::OfficialNaming::get_synonyms ( )

Undocumented method

Code:

click to view

◆ new()

public XrefMapper::OfficialNaming::new ( )

Undocumented method

Code:

click to view

◆ reset_display_xrefs()

public XrefMapper::OfficialNaming::reset_display_xrefs ( )

Undocumented method

Code:

click to view

◆ run()

public XrefMapper::OfficialNaming::run ( )

Undocumented method

Code:

click to view

sub run {
  my $self = shift;
  my $species_id = shift;
 
  my $dbname = $self->get_official_name();
  my $dbi = $self->xref->dbc;
 
  ###########################################################
  # If no offical name then we do not want to go any further
  # Just set status to official_naming_done and return
  ###########################################################
  if(!defined($dbname)){
    $self->update_process_status("official_naming_done");
    return;
  }
  $species_id = $self->get_id_from_species_name($self->core->species) unless defined $species_id;
  $self->species_id($species_id);
 
 
  ###########################################################
  # If there are any official names on transcripts or translations
  # move them onto gene level
  #
  # This is done for 2 reasons
  #  1) to make the code the same as HGNC is on a gene
  #     and it makes it easier to find.
  #  2) Later on these are copied to the canonical transcripts
  #     from the genes so move them now.
  ###########################################################
 
  if($dbname eq "MGI"){ # Copy MGI to Genes
    $self->biomart_fix("MGI","Translation","Gene");
    $self->biomart_fix("MGI","Transcript","Gene");
  }
  if($dbname eq "ZFIN_ID"){ # Copy ZFIN_ID to Genes
    $self->biomart_fix("ZFIN_ID","Translation","Gene");
    $self->biomart_fix("ZFIN_ID","Transcript","Gene");
  }
  if($dbname eq "RGD"){ # Copy RGD to Genes
    $self->biomart_fix("RGD","Translation","Gene");
    $self->biomart_fix("RGD","Transcript","Gene");
  }
 
 
 
  ######################################################
  # Get the current max values for xref and object_xref
  ######################################################
  my ($max_object_xref_id, $max_xref_id) = $self->find_max_ids($dbi);
 
  my %display_label_to_desc;
  $self->get_display_label_data(\%display_label_to_desc, $dbi);
 
  my %synonym;
  $self->get_synonyms(\%synonym, $dbi);
 
 
  # get the officail naming external_sources
  my $dbname_to_source_id = $self->get_new_dbname_sources($dbi); # reference to hash
 
  ###########################
  # Delete the old ones.
  ###########################
  $self->delete_old_data($dbname_to_source_id, $dbi);
 
  $self->reset_display_xrefs($dbi);
 
  my $db = new Bio::EnsEMBL::DBSQL::DBAdaptor(-dbconn => $self->core->dbc);
  my $ga = $db->get_GeneAdaptor();
 
  my %gene_to_transcripts;
  my %gene_id_to_stable_id;
  my %tran_id_to_stable_id;
 
  my $sql =(<<'SQ0');
SELECT gtt.gene_id, gtt.transcript_id, gsi.stable_id, tsi.stable_id 
  FROM gene_transcript_translation gtt, gene_stable_id gsi, transcript_stable_id tsi 
    WHERE gtt.gene_id = gsi.internal_id AND
          gtt.transcript_id = tsi.internal_id
    ORDER BY gsi.stable_id, tsi.stable_id
SQ0
 
  my $sth = $dbi->prepare($sql);
 
  $sth->execute;
  my ($gene_id, $tran_id, $gsi, $tsi);
  $sth->bind_columns(\$gene_id, \$tran_id, \$gsi, \$tsi);
  my @sorted_gene_ids;
  while ($sth->fetch){
    if(!defined($gene_to_transcripts{$gene_id})){
      push @sorted_gene_ids, $gene_id;
    }
    push @{$gene_to_transcripts{$gene_id}}, $tran_id;
    $gene_id_to_stable_id{$gene_id} = $gsi; 
    $tran_id_to_stable_id{$tran_id} = $tsi; 
  }
 
  my $dbentrie_sth = $self->get_dbentrie_sth($dbi);
  my $ins_xref_sth = $self->get_ins_xref_sth($dbi);
  my $ins_dep_ix_sth = $self->get_ins_dep_ix_sth($dbi);
  my $ins_object_xref_sth =  $self->get_ins_object_xref_sth($dbi);
  my $set_gene_display_xref_sth = $self->get_set_gene_display_xref_sth($dbi);
 
  my %xref_added; # store those added  $xref_added{$accession:$source_id} = $xref_id;
  my %seen_gene;
 
  my %official_name_used;
 
  my $ignore_sql =<<IEG;
  SELECT DISTINCT ox.object_xref_id
    FROM object_xref ox, dependent_xref dx,
       xref xmas, xref xdep,
       source smas, source sdep
    WHERE ox.xref_id = dx.dependent_xref_id AND
          dx.dependent_xref_id = xdep.xref_id AND
          dx.master_xref_id = xmas.xref_id AND
          xmas.source_id = smas.source_id AND
          xdep.source_id = sdep.source_id AND
          smas.name like "Refseq%predicted" AND
          sdep.name like "EntrezGene" AND
          ox.ox_status = "DUMP_OUT"
IEG
 
  my %ignore_object;
  my $ignore_sth = $dbi->prepare($ignore_sql);
  $ignore_sth->execute();
  my ($ignore_object_xref_id);
  $ignore_sth->bind_columns(\$ignore_object_xref_id);
  while($ignore_sth->fetch()){
    $ignore_object{$ignore_object_xref_id} = 1;
  }
  $ignore_sth->finish;
 
  while ( my $gene_id = shift @sorted_gene_ids){
 
    my $tran_source = $dbname;
 
    # symbols to set when found.
    my $gene_symbol = undef;
    my $gene_symbol_xref_id = undef;
    my $is_lrg = 0;
 
    ################################
    # Get offical name if it has one
    ################################
    ($gene_symbol, $gene_symbol_xref_id) = 
      $self->get_official_domain_name({gene_id       => $gene_id, 
                       gene_to_tran  => \%gene_to_transcripts,
                       gene_id_to_stable_id => \%gene_id_to_stable_id,
                                       official_name_used => \%official_name_used,
                                       dbi            => $dbi
                      });
 
    if (defined($gene_symbol_xref_id)) {
    $official_name_used{$gene_symbol_xref_id} = 1;
    }
 
    ############################################
    # If not found see if there is an LRG entry
    ############################################
    if(!defined($gene_symbol)){ # look for LRG
      ($gene_symbol, $gene_symbol_xref_id, $is_lrg) = $self->find_lrg_hgnc($gene_id, $dbi);
    }
 
    ####################################################
    # If not found look for other valid database sources
    # These are RFAM and miRBase, as well as EntrezGene
    ####################################################
    if(!defined($gene_symbol)){ 
      ($gene_symbol, $gene_symbol_xref_id) = 
      $self->find_from_other_sources(\%ignore_object, 
                                     {gene_id       => $gene_id, 
                    label_to_desc => \%display_label_to_desc,
                                      dbi           => $dbi,
                    tran_source   => \$tran_source});
    }
 
    if(defined($gene_symbol)){
      my $desc = $display_label_to_desc{$gene_symbol};
      $set_gene_display_xref_sth->execute($gene_symbol_xref_id, $gene_id);
 
      if (!$is_lrg) {
        $self->set_transcript_display_xrefs({ max_xref         => \$max_xref_id, 
                                            max_object       => \$max_object_xref_id,
                                            gene_id          =>  $gene_id,
                                            gene_id_to_stable_id => \%gene_id_to_stable_id,
                        gene_symbol      => $gene_symbol,
                        desc             => $desc, 
                                            dbi              => $dbi,
                                            source_id        => $dbname_to_source_id->{$tran_source."_trans_name"}, 
                                            xref_added       => \%xref_added, 
                                            seen_gene        => \%seen_gene, 
                                            gene_to_tran     => \%gene_to_transcripts, 
                                            tran_source      => $tran_source,
                       });
      }
    }
 
  } # for each gene
 
  $self->update_process_status('official_naming_done');
  return;
}

◆ set_the_best_odns()

public XrefMapper::OfficialNaming::set_the_best_odns ( )

Undocumented method

Code:

click to view

◆ set_transcript_display_xrefs()

public XrefMapper::OfficialNaming::set_transcript_display_xrefs ( )

Undocumented method

Code:

click to view

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

ensembl/misc-scripts/xref_mapping/XrefMapper/OfficialNaming.pm

Public Member Functions

Detailed Description

Member Function Documentation

◆ delete_old_data()

◆ find_from_other_sources()

◆ find_lrg_hgnc()

◆ find_max_ids()

◆ get_best_identity_sth()

◆ get_dbentrie_sth()

◆ get_dbentrie_with_desc_sth()

◆ get_delete_odn_sth()

◆ get_display_label_data()

◆ get_ins_dep_ix_sth()

◆ get_ins_object_xref_sth()

◆ get_ins_xref_sth()

◆ get_lrg_find_sth()

◆ get_lrg_set_status_sth()

◆ get_lrg_to_hgnc_sth()

◆ get_new_dbname_sources()

◆ get_official_domain_name()

◆ get_official_name()

◆ get_other_name_hash()

◆ get_set_gene_display_xref_sth()

◆ get_set_transcript_display_xref_sth()

◆ get_synonyms()

◆ new()

◆ reset_display_xrefs()

◆ run()

◆ set_the_best_odns()

◆ set_transcript_display_xrefs()