Functions
public void	parse_deleted_files ()

public void	determine_deleted ()

public Int	create_mapping_session ()

public void	create_stable_id_events ()

public void	increment_gene_versions ()

public void	mark_deleted ()

public	populated_archive ()

Function Documentation

◆ create_mapping_session()

public Int create_mapping_session ( )

  Example     :

my $msi = &create_mapping_session;

  Description : Creates a new mapping_session in the db.
  Return type : Int - mapping_session_id of the newly created entry
  Exceptions  : none
  Caller      : main()
  Status      : At Risk
              : under development

Code:

click to view

◆ create_stable_id_events()

public void create_stable_id_events ( )

  Example     :

&create_stable_id_events

  Description : Creates stable_id_event entries for all objects found in the old
                db, mapping them to themselves. Optionally, some biotypes will
                be skipped (this is useful if a separate script is run to deal
                with ncRNAs).
  Return type : none
  Exceptions  : none
  Caller      : main()
  Status      : At Risk
              : under development

Code:

click to view

◆ determine_deleted()

public void determine_deleted ( )

  Example     :

&determine_deleted;

  Description : Infer deleted genes/transcripts from dbs by comparing which
                objects are in the old and new db.
  Return type : none
  Exceptions  : none
  Caller      : main()
  Status      : At Risk
              : under development

Code:

click to view

sub determine_deleted {
  
  $support->log_stamped("Determining list of deleted gene, transcript and translation stable IDs by comparing dbs...\n");
 
  # optionally skip ncRNAs
  #
  # this is the complete list of ncRNA biotype; you might need to update it (the
  # code below will try to help you with this)
  my @nc_biotypes = qw(
    miRNA
    miRNA_pseudogene
    misc_RNA
    misc_RNA_pseudogene
    Mt_rRNA
    Mt_tRNA
    Mt_tRNA_pseudogene
    rRNA
    rRNA_pseudogene
    scRNA
    scRNA_pseudogene
    snoRNA
    snoRNA_pseudogene
    snRNA
    snRNA_pseudogene
    tRNA_pseudogene
  );
 
  if ($support->param('skip_ncrna')) {
 
    %skip_biotypes = map { $_ => 1 } @nc_biotypes;
 
    # make sure we have a complete list of ncRNA biotypes
    my $sql = qq(SELECT DISTINCT biotype from gene);
    my $sth1 = $dbh_new->prepare($sql);
    $sth1->execute;
    my @biotypes_db;
    
    while ((my $biotype) = $sth1->fetchrow_array) {
      push @biotypes_db, $biotype unless ($skip_biotypes{$biotype});
    }
 
    $sth1->finish;
 
    if (@biotypes_db) {
      print "These are the non-ncRNA biotypes found in the db:\n";
      map { print "  $_\n" } @biotypes_db;
      print "\nPlease check that the list of ncRNA biotypes is still complete, otherwise adapt the script.\n";
      exit unless $support->user_proceed("Continue?");
    }
  }
 
  # optionally skip other biotypes
  if ($support->param('skip_biotypes')) {
    %skip_biotypes = map { $_ => 1 } $support->param('skip_biotypes');
  }
 
  # get old and new genes and transcripts from db
  my $ga_old = $dba_old->get_GeneAdaptor;
  my @genes_old = @{ $ga_old->fetch_all };
 
  my $ga_new = $dba_new->get_GeneAdaptor;
  my %genes_new = map { $_->stable_id => $_ } @{ $ga_new->fetch_all };
  my $ta_new = $dba_new->get_TranscriptAdaptor;
  my %tsi_new = map { $_ => 1 } @{ $ta_new->list_stable_ids };
 
  while (my $g_old = shift(@genes_old)) {
    
    # skip biotypes
    next if ($skip_biotypes{$g_old->biotype});
  
    my $gsi = $g_old->stable_id;
  
    # mark gene as deleted
    unless ($genes_new{$gsi}) {
      $gsi_del{$gsi} = 1;
      $genes{$gsi} = $g_old;
    }
 
    # transcripts
    foreach my $tr (@{ $g_old->get_all_Transcripts }) {
 
      # skip biotypes
      next if ($skip_biotypes{$tr->biotype});
  
      my $tsi = $tr->stable_id;
 
      unless ($tsi_new{$tsi}) {
        # mark transcript and translation as deleted
        $tsi_del{$tsi} = 1;
        my $tl = $tr->translation;
        $tlsi_del{$tl->stable_id} = 1 if ($tl);
 
        # mark gene as modified
        $gsi_mod{$gsi} = 1;
        $genes_mod{$gsi} = $g_old;
        $genes{$gsi} = $g_old;
      }
    }
  }
 
  # create stable ID strings for use in mysql IN statements
  $gsi_string = "'".join("', '", keys(%gsi_del))."'";
  $gsi_mod_string = "'".join("', '", keys(%gsi_mod))."'";
  $tsi_string = "'".join("', '", keys(%tsi_del))."'";
  $tlsi_string = "'".join("', '", keys(%tlsi_del))."'";
 
  # stats
  my $fmt = "%-15s%6d\n";
  $support->log("Deleted objects found:\n", 1);
  $support->log(sprintf($fmt, "genes", scalar(keys(%gsi_del))), 2);
  $support->log(sprintf($fmt, "transcripts", scalar(keys(%tsi_del))), 2);
  $support->log(sprintf($fmt, "translations", scalar(keys(%tlsi_del))), 2);
  
  $support->log("Modified genes found: ".scalar(keys(%gsi_mod))."\n", 1);
 
  $support->log_stamped("Done.\n\n");
}

◆ increment_gene_versions()

public void increment_gene_versions ( )

  Example     :

&increment_gene_versions;

  Description : Increment version of all genes where transcripts were deleted.
                Also checks that gene_stable_id.stable_id is correct (and adjusts
                if necessary).
  Return type : none
  Exceptions  : none
  Caller      : main()
  Status      : At Risk
              : under development

Code:

click to view

◆ mark_deleted()

public void mark_deleted ( )

  Example     :

&mark_deleted;

  Description : Sets stable_id_event.new_stable_id to NULL for deleted objects.
  Return type : none
  Exceptions  : none
  Caller      : main()
  Status      : At Risk
              : under development

Code:

click to view

◆ parse_deleted_files()

public void parse_deleted_files ( )

  Example     :

&parse_deleted_files;

  Description : Deprecated:

                Read list of deleted gene and transcript stable IDs from file(s).
                Note that this method of determining which objects were deleted
                is now 
(because it was error-prone when dealing with 
                both whole gene and individual transcript deletions).
  Return type : none
  Exceptions  : thrown on missing files
  Caller      : main()
  Status      : At Risk
              : under development

Code:

click to view

sub parse_deleted_files {
 
  $support->log_warning("DEPRECATED. Don't use stable ID files (this is error-prone), rather let the script determine which objects were deleted from the dbs.\n");
 
  my $ta = $dba_old->get_TranscriptAdaptor;
  my $ga = $dba_old->get_GeneAdaptor;
 
  #
  # read list of deleted gene_stable_ids from file
  #
  $support->log_stamped("Reading list of deleted gene_stable_ids from file, and fetching associated transcript and translation stable IDs from the db...\n");
  my $gfh = $support->filehandle('<', $support->param('gene_stable_id_file'));
 
  while (my $g = <$gfh>) {
    chomp $g;
    my $gene = $ga->fetch_by_stable_id($g);
 
    $genes{$g} = $gene;
    $gsi_del{$g} = 1;
 
    # fetch associated transcript and translation stable IDs from the old db
    foreach my $transcript (@{ $gene->get_all_Transcripts }) {
      $tsi_del{$transcript->stable_id} = 1;
      my $tl = $transcript->translation;
      $tlsi_del{$tl->stable_id} = 1 if ($tl);
    }
  }
 
  #
  # read list of deleted transcript_stable_ids from file
  #
  if ($support->param('transcript_stable_id_file')) {
 
    $support->log_stamped("Reading list of deleted transcript_stable_ids from file, and fetching associated translation stable IDs from the db...\n");
 
    my $tfh = $support->filehandle('<', $support->param('transcript_stable_id_file'));
 
    while (my $t = <$tfh>) {
      chomp $t;
      my $transcript = $ta->fetch_by_stable_id($t);
 
      # skip non-protein-coding genes
      unless ($transcript->biotype eq 'protein_coding') {
        $support->log_warning("Transcript ".$transcript->stable_id." is non-protein_coding, skipping.\n", 1);
        next;
      }
      
      $tsi_del{$transcript->stable_id} = 1;
      my $tl = $transcript->translation;
      $tlsi_del{$tl->stable_id} = 1 if ($tl);
      
      my $gene = $ga->fetch_by_transcript_id($transcript->dbID);
      $genes{$gene->stable_id} = $gene;
      $gsi_mod{$gene->stable_id} = 1;
    }
  }
 
  $gsi_string = "'".join("', '", keys(%gsi_del))."'";
  $gsi_mod_string = "'".join("', '", keys(%gsi_mod))."'";
  $tsi_string = "'".join("', '", keys(%tsi_del))."'";
  $tlsi_string = "'".join("', '", keys(%tlsi_del))."'";
 
  $support->log_stamped("Done loading ".scalar(keys(%gsi_del))." gene, ".scalar(keys(%tsi_del))." transcript and ".scalar(keys(%tlsi_del))." translation stable IDs.\n\n");
 
}

◆ populated_archive()

public populated_archive ( )

Undocumented method

Code:

click to view

Functions

Function Documentation

◆ create_mapping_session()

◆ create_stable_id_events()

◆ determine_deleted()

◆ increment_gene_versions()

◆ mark_deleted()

◆ parse_deleted_files()

◆ populated_archive()