Public Member Functions
public	new ()

public	force ()

public	dbh ()

public	user ()

public	host ()

public	port ()

public	password ()

public	verbose ()

public	schema ()

public	vegaschema ()

public	source ()

public	target ()

public	limit ()

public	debug ()

public	copy_tables ()

public	get_coord_system_id ()

public	get_default_assembly ()

public	contig_to_seq_region ()

public	clone_to_seq_region ()

public	chromosome_to_seq_region ()

public	supercontig_to_seq_region ()

public	assembly_contig_chromosome ()

public	assembly_contig_clone ()

public	assembly_contig_supercontig ()

public	assembly_supercontig_chromosome ()

public	create_coord_systems ()

public	create_meta_coord ()

public	create_seq_regions ()

public	create_assembly ()

public	transfer_dna ()

public	transfer_genes ()

public	transfer_prediction_transcripts ()

public	transfer_features ()

public	transfer_stable_ids ()

public	transfer_vega_stable_ids ()

public	transfer_meta ()

public	copy_other_tables ()

public	copy_repeat_consensus ()

public	create_attribs ()

public	set_top_level ()

public	add_attrib_code ()

Detailed Description

Definition at line 3 of file BasicConverter.pm.

Member Function Documentation

◆ add_attrib_code()

public SeqStoreConverter::BasicConverter::add_attrib_code ( )

Undocumented method

Code:

click to view

◆ assembly_contig_chromosome()

public SeqStoreConverter::BasicConverter::assembly_contig_chromosome ( )

Undocumented method

Code:

click to view

◆ assembly_contig_clone()

public SeqStoreConverter::BasicConverter::assembly_contig_clone ( )

Undocumented method

Code:

click to view

◆ assembly_contig_supercontig()

public SeqStoreConverter::BasicConverter::assembly_contig_supercontig ( )

Undocumented method

Code:

click to view

◆ assembly_supercontig_chromosome()

public SeqStoreConverter::BasicConverter::assembly_supercontig_chromosome ( )

Undocumented method

Code:

click to view

◆ chromosome_to_seq_region()

public SeqStoreConverter::BasicConverter::chromosome_to_seq_region ( )

Undocumented method

Code:

click to view

◆ clone_to_seq_region()

public SeqStoreConverter::BasicConverter::clone_to_seq_region ( )

Undocumented method

Code:

click to view

◆ contig_to_seq_region()

public SeqStoreConverter::BasicConverter::contig_to_seq_region ( )

Undocumented method

Code:

click to view

◆ copy_other_tables()

public SeqStoreConverter::BasicConverter::copy_other_tables ( )

Undocumented method

Code:

click to view

◆ copy_repeat_consensus()

public SeqStoreConverter::BasicConverter::copy_repeat_consensus ( )

Undocumented method

Code:

click to view

◆ copy_tables()

public SeqStoreConverter::BasicConverter::copy_tables ( )

Undocumented method

Code:

click to view

◆ create_assembly()

public SeqStoreConverter::BasicConverter::create_assembly ( )

Undocumented method

Code:

click to view

◆ create_attribs()

public SeqStoreConverter::BasicConverter::create_attribs ( )

Undocumented method

Code:

click to view

◆ create_coord_systems()

public SeqStoreConverter::BasicConverter::create_coord_systems ( )

Undocumented method

Code:

click to view

◆ create_meta_coord()

public SeqStoreConverter::BasicConverter::create_meta_coord ( )

Undocumented method

Code:

click to view

◆ create_seq_regions()

public SeqStoreConverter::BasicConverter::create_seq_regions ( )

Undocumented method

Code:

click to view

◆ dbh()

public SeqStoreConverter::BasicConverter::dbh ( )

Undocumented method

Code:

click to view

◆ debug()

public SeqStoreConverter::BasicConverter::debug ( )

Undocumented method

Code:

click to view

◆ force()

public SeqStoreConverter::BasicConverter::force ( )

Undocumented method

Code:

click to view

◆ get_coord_system_id()

public SeqStoreConverter::BasicConverter::get_coord_system_id ( )

Undocumented method

Code:

click to view

◆ get_default_assembly()

public SeqStoreConverter::BasicConverter::get_default_assembly ( )

Undocumented method

Code:

click to view

◆ host()

public SeqStoreConverter::BasicConverter::host ( )

Undocumented method

Code:

click to view

◆ limit()

public SeqStoreConverter::BasicConverter::limit ( )

Undocumented method

Code:

click to view

◆ new()

public SeqStoreConverter::BasicConverter::new ( )

Undocumented method

Code:

click to view

sub new {
  my ( $class, $user, $pass, $host, $source, $target, $schema, $vega_schema, $force, $verbose, $limit ) = @_;
 
  my $self = bless {}, $class;
 
  my $port;
  ($host, $port) = split(/:/, $host);
  $port ||= 3306;
 
  my $dbh = DBI->connect( "DBI:mysql:host=$host:port=$port", $user, $pass,
                          {'RaiseError' => 1});
 
 
  $self->verbose( $verbose );
  $self->dbh( $dbh );
  $self->force( $force );
  $self->source( $source );
  $self->target( $target );
  $self->schema( $schema );
  $self->vegaschema( $vega_schema);
  $self->host( $host );
  $self->password( $pass);
  $self->user($user);
  $self->port($port);
  $self->limit($limit);
 
 
  #check to see if the destination and source databases exist already.
  my %dbs = map {$_->[0] => 1} @{$dbh->selectall_arrayref('show databases')};
  if( !$dbs{$source} ) {
    die ("Source db $source does not exist" );
  }
 
  if( $dbs{$target} ) {
    if( $force ) {
      $dbh->do( "drop database $target" );
    } else {
      die("Target db $target already exists. Use -force option to overwrite.");
    }
  }
 
  $dbh->do( "create database ".$self->target() );
 
  $self->debug("Building schema for $target from $schema");
  die "Cannot open $schema" if (! -e $schema);
  my $cmd = "/usr/local/mysql/bin/mysql -u $user -p$pass -P $port -h $host $target < $schema";
  system ($cmd);
 
  if ($vega_schema) {
      $self->debug("Adding vega tables for $target");
      die "Cannot open vega creation script" if (! -e $vega_schema);
      my $cmd = "/usr/local/mysql/bin/mysql -u $user -p$pass -P $port -h $host $target < $vega_schema";
      system ($cmd);
  }
 
  $self->debug("Creating temporary tables");
  #create a temporary table to store the mapping of old ids to new ids
  $dbh->do
    ("CREATE TEMPORARY TABLE $target.tmp_cln_map (" .
     "old_id INT, new_id INT, INDEX new_idx (new_id))");
 
  #create a temp table which will store the mapping of old chromosome
  #identifiers to new identifiers
  $dbh->do("CREATE TEMPORARY TABLE $target.tmp_chr_map (" .
           "  old_id INT, new_id INT,".
           "  INDEX new_idx (new_id))");
 
  #create a temporary table to hold old supercontig name -> new id mappings
  $dbh->do("CREATE TEMPORARY TABLE $target.tmp_superctg_map (" .
           "name VARCHAR(255), new_id INT, ".
           "INDEX new_idx (new_id))");
 
  return $self;
}

◆ password()

public SeqStoreConverter::BasicConverter::password ( )

Undocumented method

Code:

click to view

◆ port()

public SeqStoreConverter::BasicConverter::port ( )

Undocumented method

Code:

click to view

◆ schema()

public SeqStoreConverter::BasicConverter::schema ( )

Undocumented method

Code:

click to view

◆ set_top_level()

public SeqStoreConverter::BasicConverter::set_top_level ( )

Undocumented method

Code:

click to view

◆ source()

public SeqStoreConverter::BasicConverter::source ( )

Undocumented method

Code:

click to view

◆ supercontig_to_seq_region()

public SeqStoreConverter::BasicConverter::supercontig_to_seq_region ( )

Undocumented method

Code:

click to view

◆ target()

public SeqStoreConverter::BasicConverter::target ( )

Undocumented method

Code:

click to view

◆ transfer_dna()

public SeqStoreConverter::BasicConverter::transfer_dna ( )

Undocumented method

Code:

click to view

◆ transfer_features()

public SeqStoreConverter::BasicConverter::transfer_features ( )

Undocumented method

Code:

click to view

sub transfer_features {
  my $self = shift;
 
  my $target = $self->target();
  my $source = $self->source();
  my $dbh    = $self->dbh();
 
  my $limit = '';
  if($self->limit()) {
    $limit = ' limit ' . $self->limit();
  }
 
 
  #
  # Feature tables
  # Note that we can just rename contig_* to set_region_* since the
  # contig IDs were copied verbatim into seq_region
  #
 
  # For some reason mysql occasionally refuses to use the index on large 
  # tables following copies like the following.
  # So: drop the indexes first and then re-add them after
 
  # simple_feature
  $self->debug("Building simple_feature table");
  $dbh->do
    ("INSERT INTO $target.simple_feature (simple_feature_id, seq_region_id, ".
     "       seq_region_start, seq_region_end, seq_region_strand, " .
     "       display_label, analysis_id, score) " .
     "SELECT simple_feature_id, contig_id, contig_start, contig_end, " .
     "       contig_strand, display_label, analysis_id, score " .
     "FROM $source.simple_feature $limit");
  
  # repeat_feature
  $self->debug("Dropping indexes on repeat_feature");
  $dbh->do("ALTER TABLE $target.repeat_feature DROP INDEX seq_region_idx");
  $dbh->do("ALTER TABLE $target.repeat_feature DROP INDEX repeat_idx");
  $dbh->do("ALTER TABLE $target.repeat_feature DROP INDEX analysis_idx");
 
  $self->debug("Building repeat_feature table");
  $dbh->do
    ("INSERT INTO $target.repeat_feature (repeat_feature_id, seq_region_id, " .
     "    seq_region_start, seq_region_end, seq_region_strand, analysis_id, " .
     "    repeat_start, repeat_end, repeat_consensus_id, score) " .
     "SELECT repeat_feature_id, contig_id, contig_start, contig_end, " .
     "       contig_strand, analysis_id, repeat_start, repeat_end, " .
     "       repeat_consensus_id, score FROM $source.repeat_feature $limit");
 
  $self->debug("Reading indexes on repeat_feature");
  $dbh->do("ALTER TABLE $target.repeat_feature " .
           "ADD INDEX seq_region_idx( seq_region_id, seq_region_start)");
  $dbh->do("ALTER TABLE $target.repeat_feature " .
           "ADD INDEX repeat_idx( repeat_consensus_id )");
  $dbh->do("ALTER TABLE $target.repeat_feature " .
           "ADD INDEX analysis_idx(analysis_id)");
 
  # protein_align_feature
  $self->debug("Dropping indexes on protein_align_feature");
  $dbh->do("ALTER TABLE $target.protein_align_feature DROP INDEX hit_idx");
  $dbh->do( "ALTER TABLE $target.protein_align_feature " .
            "DROP INDEX seq_region_idx");
 
  $self->debug("Building protein_align_feature table");
  $dbh->do( "INSERT INTO $target.protein_align_feature " .
            "  (protein_align_feature_id, seq_region_id, seq_region_start, " .
            "   seq_region_end, seq_region_strand, analysis_id, hit_start, " .
            "   hit_end, hit_name, cigar_line, evalue, perc_ident, score) " .
            "SELECT protein_align_feature_id, contig_id, contig_start, " .
            "       contig_end, contig_strand, analysis_id, hit_start, " .
            "       hit_end, hit_name, cigar_line, evalue, perc_ident, score ".
            "FROM $source.protein_align_feature $limit");
 
  $self->debug("Reading indexes on protein_align_feature");
  $dbh->do( qq{ALTER TABLE $target.protein_align_feature
               ADD index  seq_region_idx(  analysis_id, seq_region_id,
                                           seq_region_start, score )});
  $dbh->do( "ALTER TABLE $target.protein_align_feature " .
            "ADD index hit_idx(hit_name)");
 
  # dna_align_feature
  $self->debug("Dropping indexes on dna_align_feature");
  $dbh->do( "ALTER TABLE $target.dna_align_feature DROP INDEX seq_region_idx");
  $dbh->do( "ALTER TABLE $target.dna_align_feature DROP INDEX hit_idx");
 
  $self->debug("Building dna_align_feature table");
  $dbh->do( "INSERT INTO $target.dna_align_feature " .
            "       (dna_align_feature_id, seq_region_id, seq_region_start, ".
            "        seq_region_end, seq_region_strand, analysis_id, " .
            "        hit_start, hit_end, hit_name, hit_strand, cigar_line, " .
            "        evalue, perc_ident, score) " .
            "SELECT dna_align_feature_id, contig_id, contig_start, " .
            "       contig_end, contig_strand, analysis_id, hit_start, " .
            "       hit_end, hit_name, hit_strand, cigar_line, evalue, " .
            "       perc_ident, score FROM $source.dna_align_feature $limit");
 
 
  $self->debug("Reading indexes on dna_align_feature");
  $dbh->do( qq{ALTER TABLE $target.dna_align_feature
               ADD INDEX seq_region_idx(seq_region_id, analysis_id,
                                        seq_region_start, score)});
  $dbh->do( "ALTER TABLE $target.dna_align_feature " .
            "ADD index hit_idx(hit_name)");
 
  # marker_feature
  $self->debug("Building marker_feature table");
  $dbh->do( "INSERT INTO $target.marker_feature " .
            "       (marker_feature_id, marker_id, seq_region_id, " .
            "        seq_region_start, seq_region_end, analysis_id, " .
            "        map_weight) " .
            "SELECT marker_feature_id, marker_id, contig_id, contig_start, ".
            "       contig_end, analysis_id, map_weight " .
            "FROM   $source.marker_feature $limit");
  
  # qtl_feature
  # Note this uses chromosome coords so we have to join with tmp_chr_map to 
  # get the mapping
  $self->debug("Building qtl_feature table");
 
  $dbh->do
    ("INSERT INTO $target.qtl_feature( seq_region_id, seq_region_start, " .
     "  seq_region_end, qtl_id, analysis_id) " .
     "SELECT tcm.new_id, " .
     "       q.start, q.end, q.qtl_id, q.analysis_id " .
     "FROM $target.tmp_chr_map tcm, $source.qtl_feature q " .
     "WHERE tcm.old_id = q.chromosome_id $limit");
  
  # These tables now have seq_region_* instead of chromosome_*
  
  $self->debug("Building karyotype table");
  $dbh->do(
           "INSERT INTO $target.karyotype " .
           "SELECT null, tcm.new_id, " .
           "       k.chr_start, k.chr_end, k.band, k.stain " .
           "FROM $target.tmp_chr_map tcm, $source.karyotype k " .
           "WHERE tcm.old_id = k.chromosome_id $limit");
 
 
  $self->debug("Building marker_map_location table");
  $dbh->do(
           "INSERT INTO $target.marker_map_location " .
           "SELECT mml.marker_id, mml.map_id, " .
           "       c.name, " .
           "       mml.marker_synonym_id, mml.position, mml.lod_score " .
           "FROM $source.chromosome c, $source.marker_map_location mml " .
           "WHERE c.chromosome_id = mml.chromosome_id $limit");
 
  $self->debug( "Building misc_feature table" );
  $dbh->do
    ("INSERT INTO $target.misc_feature( misc_feature_id, seq_region_id, " . 
     "            seq_region_start, seq_region_end, seq_region_strand ) " .
     "SELECT m.mapfrag_id, sr.seq_region_id, m.seq_start, m.seq_end, " .
     "       m.orientation " .
     "FROM   $source.mapfrag m, $target.seq_region sr, $source.dnafrag d " . 
     "WHERE  m.dnafrag_id = d.dnafrag_id " .
     "AND    d.name = sr.name $limit" );
 
  $self->debug( "Building misc_set table" );
  $dbh->do
    ("INSERT INTO $target.misc_set( misc_set_id, code, name, description, " .
     "                              max_length ) " .
     "SELECT mapset_id, code, name, description, max_length " . 
     "FROM $source.mapset ms" );
 
  $self->debug( "Building misc_attrib table" );
  $dbh->do
     ("INSERT INTO $target.misc_attrib( misc_feature_id, attrib_type_id, " . 
    "                    value ) ". 
    "SELECT mapfrag_id, mapannotationtype_id, value " .
    "FROM $source.mapannotation" );
 
  $dbh->do
   ("INSERT INTO $target.misc_attrib( misc_feature_id, attrib_type_id, " .
    "                                 value) " .
    "SELECT mf.mapfrag_id, at.attrib_type_id, mf.name " .
    "FROM   $source.mapfrag mf, $target.attrib_type at " .
    "WHERE  at.code = 'name'");
 
   $dbh->do
   ("INSERT INTO $target.misc_attrib( misc_feature_id, attrib_type_id, " .
    "                                 value) " .
    "SELECT mf.mapfrag_id, at.attrib_type_id, mf.type " .
    "FROM   $source.mapfrag mf, $target.attrib_type at " .
    "WHERE  at.code = 'type'");
 
  $self->debug( "Building misc_feature_misc_set table" );
  $dbh->do
    ("INSERT INTO $target.misc_feature_misc_set(misc_feature_id, misc_set_id)".
     "SELECT mapfrag_id, mapset_id ".
     "FROM $source.mapfrag_mapset $limit" );
 
  return;
}

◆ transfer_genes()

public SeqStoreConverter::BasicConverter::transfer_genes ( )

Undocumented method

Code:

click to view

sub transfer_genes {
  my $self = shift;
 
  my $target = $self->target();
  my $source = $self->source();
  my $dbh    = $self->dbh();
 
  #
  # Transfer the gene table
  #
 
  $self->debug("Building gene table");
 
  $dbh->do
    ("INSERT INTO $target.gene " .
     "SELECT g.gene_id, g.type, g.analysis_id, tcm.new_id, " .
     "MIN(IF (a.contig_ori=1,(e.contig_start+a.chr_start-a.contig_start)," .
     "       (a.chr_start+a.contig_end-e.contig_end ))) as start, " .
     "MAX(IF (a.contig_ori=1,(e.contig_end+a.chr_start-a.contig_start), " .
     "       (a.chr_start+a.contig_end-e.contig_start))) as end, " .
     "       a.contig_ori*e.contig_strand as strand, " .
     "       g.display_xref_id " .
     "FROM   $source.transcript t, $source.exon_transcript et, " .
     "       $source.exon e, $source.assembly a, $source.gene g, " .
     "       $target.tmp_chr_map tcm " .
     "WHERE  t.transcript_id = et.transcript_id " .
     "AND    et.exon_id = e.exon_id " .
     "AND    e.contig_id = a.contig_id " .
     "AND    g.gene_id = t.gene_id " .
     "AND    a.chromosome_id = tcm.old_id " .
     "GROUP BY g.gene_id");
 
 
  # 
  # Transfer the transcript table
  #
  $self->debug("Building transcript table");
 
  $dbh->do
    ("INSERT INTO $target.transcript " .
     "SELECT t.transcript_id, t.gene_id, tcm.new_id, " .
     "MIN(IF (a.contig_ori=1,(e.contig_start+a.chr_start-a.contig_start)," .
     "       (a.chr_start+a.contig_end-e.contig_end ))) as start, " .
     "MAX(IF (a.contig_ori=1,(e.contig_end+a.chr_start-a.contig_start), " .
     "       (a.chr_start+a.contig_end-e.contig_start))) as end, " .
     "       a.contig_ori*e.contig_strand as strand, " .
     "       t.display_xref_id " .
     "FROM   $source.transcript t, $source.exon_transcript et, " .
     "       $source.exon e, $source.assembly a, $target.tmp_chr_map tcm " .
     "WHERE  t.transcript_id = et.transcript_id " .
     "AND    et.exon_id = e.exon_id " .
     "AND    e.contig_id = a.contig_id " .
     "AND    a.chromosome_id = tcm.old_id " .
     "GROUP BY t.transcript_id");
 
  #
  # Transfer the exon table
  #
  $self->debug("Building exon table");
 
  $dbh->do
    ("INSERT INTO $target.exon " .
     "SELECT e.exon_id, tcm.new_id, " .
     "MIN(IF (a.contig_ori=1,(e.contig_start+a.chr_start-a.contig_start)," .
     "       (a.chr_start+a.contig_end-e.contig_end ))) as start, " .
     "MAX(IF (a.contig_ori=1,(e.contig_end+a.chr_start-a.contig_start), " .
     "       (a.chr_start+a.contig_end-e.contig_start))) as end, " .
     "       a.contig_ori*e.contig_strand as strand, " .
     "       e.phase, e.end_phase " .
     "FROM   $source.transcript t, $source.exon_transcript et, " .
     "       $source.exon e, $source.assembly a, $source.gene g, " .
     "       $target.tmp_chr_map tcm " .
     "WHERE  t.transcript_id = et.transcript_id " .
     "AND    et.exon_id = e.exon_id " .
     "AND    e.contig_id = a.contig_id " .
     "AND    g.gene_id = t.gene_id " .
     "AND    a.chromosome_id = tcm.old_id " .
     "GROUP BY e.exon_id");
 
  # 
  # Transfer translation table
  # 
 
  $self->debug("Building translation table");
 
  $dbh->do
    ("INSERT INTO $target.translation " .
     "SELECT tl.translation_id, ts.transcript_id, tl.seq_start, " .
     "       tl.start_exon_id, tl.seq_end, tl.end_exon_id " .
     "FROM $source.transcript ts, $source.translation tl " .
     "WHERE ts.translation_id = tl.translation_id");
 
  return;
}

◆ transfer_meta()

public SeqStoreConverter::BasicConverter::transfer_meta ( )

Undocumented method

Code:

click to view

◆ transfer_prediction_transcripts()

public SeqStoreConverter::BasicConverter::transfer_prediction_transcripts ( )

Undocumented method

Code:

click to view

◆ transfer_stable_ids()

public SeqStoreConverter::BasicConverter::transfer_stable_ids ( )

Undocumented method

Code:

click to view

◆ transfer_vega_stable_ids()

public SeqStoreConverter::BasicConverter::transfer_vega_stable_ids ( )

Undocumented method

Code:

click to view

◆ user()

public SeqStoreConverter::BasicConverter::user ( )

Undocumented method

Code:

click to view

◆ vegaschema()

public SeqStoreConverter::BasicConverter::vegaschema ( )

Undocumented method

Code:

click to view

◆ verbose()

public SeqStoreConverter::BasicConverter::verbose ( )

Undocumented method

Code:

click to view

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

ensembl/misc-scripts/surgery/SeqStoreConverter/BasicConverter.pm

Public Member Functions

Detailed Description

Member Function Documentation

◆ add_attrib_code()

◆ assembly_contig_chromosome()

◆ assembly_contig_clone()

◆ assembly_contig_supercontig()

◆ assembly_supercontig_chromosome()

◆ chromosome_to_seq_region()

◆ clone_to_seq_region()

◆ contig_to_seq_region()

◆ copy_other_tables()

◆ copy_repeat_consensus()

◆ copy_tables()

◆ create_assembly()

◆ create_attribs()

◆ create_coord_systems()

◆ create_meta_coord()

◆ create_seq_regions()

◆ dbh()

◆ debug()

◆ force()

◆ get_coord_system_id()

◆ get_default_assembly()

◆ host()

◆ limit()

◆ new()

◆ password()

◆ port()

◆ schema()

◆ set_top_level()

◆ source()

◆ supercontig_to_seq_region()

◆ target()

◆ transfer_dna()

◆ transfer_features()

◆ transfer_genes()

◆ transfer_meta()

◆ transfer_prediction_transcripts()

◆ transfer_stable_ids()

◆ transfer_vega_stable_ids()

◆ user()

◆ vegaschema()

◆ verbose()