Skip to content
Snippets Groups Projects
DBEntryAdaptor.pm 68.3 KiB
Newer Older
Andy Yates's avatar
Andy Yates committed
  Copyright (c) 1999-2012 The European Bioinformatics Institute and
  Genome Research Limited.  All rights reserved.

  This software is distributed under a modified Apache license.
  For license details, please see

    http://www.ensembl.org/info/about/code_licence.html

=head1 CONTACT

  Please email comments or questions to the public Ensembl

  Questions may also be sent to the Ensembl help desk at
  <helpdesk@ensembl.org>.

=cut
Bio::EnsEMBL::DBSQL::DBEntryAdaptor -
MySQL Database queries to load and store external object references.

=head1 SYNOPSIS

  $db_entry_adaptor =
    $registry->get_adaptor( 'Human', 'Core', 'DBEntry' );
  $db_entry = $db_entry_adaptor->fetch_by_dbID($id);
  my $gene_adaptor = $registry->get_adaptor( 'Human', 'Core', 'Gene' );

  my $gene = $gene_adaptor->fetch_by_stable_id('ENSG00000101367');

  @db_entries = @{ $db_entry_adaptor->fetch_all_by_Gene($gene) };
  @gene_ids   = $db_entry_adaptor->list_gene_ids_by_extids('BAB15482');
Graham McVicker's avatar
Graham McVicker committed

=cut

package Bio::EnsEMBL::DBSQL::DBEntryAdaptor;

use Bio::EnsEMBL::DBSQL::BaseAdaptor;
use Bio::EnsEMBL::DBEntry;
use Bio::EnsEMBL::Utils::Exception qw(deprecate throw warning);
use vars qw(@ISA);
use strict;

@ISA = qw( Bio::EnsEMBL::DBSQL::BaseAdaptor );

Graham McVicker's avatar
Graham McVicker committed
=head2 fetch_by_dbID

  Arg [1]    : int $dbID
               the unique database identifier for the DBEntry to retrieve
  Example    : my $db_entry = $db_entry_adaptor->fetch_by_dbID($dbID);
  Description: Retrieves a dbEntry from the database via its unique
               identifier.
Graham McVicker's avatar
Graham McVicker committed
  Returntype : Bio::EnsEMBL::DBEntry
  Exceptions : none
  Caller     : general
Graham McVicker's avatar
Graham McVicker committed

=cut

sub fetch_by_dbID {
  my $sth = $self->prepare(
    "SELECT  xref.xref_id,
            xref.dbprimary_acc,
            xref.display_label,
            xref.version,
            exDB.priority,
            exDB.db_name,
            exDB.db_display_name,
            exDB.db_release,
            es.synonym,
            xref.info_type,
            xref.info_text,
            exDB.type,
            exDB.secondary_db_name,
            xref.description
    FROM    (xref, external_db exDB)
    LEFT JOIN external_synonym es ON
            es.xref_id = xref.xref_id
    WHERE   xref.xref_id = ?
    AND     xref.external_db_id = exDB.external_db_id" );

  $sth->bind_param( 1, $dbID, SQL_INTEGER );
  my $exDB;
  my $max_rows = 1000;

  while ( my $rowcache = $sth->fetchall_arrayref( undef, $max_rows ) ) {
      #$description refers to the external_db description, while $desc was referring the xref description
    while ( my $arrayref = shift( @{$rowcache} ) ) {
      my ( $refID,               $dbprimaryId,
           $displayid,           $version,
           $dbname,              $db_display_name,
           $release,             $synonym,
           $info_type,           $info_text,
           $type,                $secondary_db_name,
      ) = @$arrayref;

      if ( !defined($exDB) ) {
        $exDB =
          Bio::EnsEMBL::DBEntry->new(
                           -adaptor             => $self,
                           -dbID                => $dbID,
                           -primary_id          => $dbprimaryId,
                           -display_id          => $displayid,
                           -version             => $version,
                           -release             => $release,
                           -dbname              => $dbname,
                           -priority            => $priority,
                           -db_display_name     => $db_display_name,
                           -info_type           => $info_type,
                           -info_text           => $info_text,
                           -type                => $type,
                           -secondary_db_name   => $secondary_db_name,
                           -secondary_db_table  => $secondary_db_table,
      if ( defined($synonym) ) { $exDB->add_synonym($synonym) }
    } ## end while ( my $arrayref = shift...
  } ## end while ( my $rowcache = $sth...

  $sth->finish();

  return $exDB;
sub _get_all_dm_loc_sth {
  my ($self, $constraint ,$ensembl_object ) = @_;
  my $object_type;
  if($ensembl_object->isa("Bio::EnsEMBL::Gene")){
    $object_type = "Gene";
  }
  elsif($ensembl_object->isa("Bio::EnsEMBL::Transcript")){
    $object_type = "Transcript";
  }
  elsif($ensembl_object->isa("Bio::EnsEMBL::Translation")){
    $object_type = "Translation";
  }
  elsif($ensembl_object->isa("Bio::EnsEMBL::Operon")){
    $object_type = "Operon";
  }
  elsif($ensembl_object->isa("Bio::EnsEMBL::OperonTranscript")){
    $object_type = "OperonTranscript";
  }
  else{
    warn(ref($ensembl_object)." is not a Gene Transcript or Translation object??\n");
    return undef;
  }
  my $sql = "SELECT xref.xref_id,
               xref.dbprimary_acc,
            xref.display_label,
            xref.version,
            exDB.priority,
            exDB.db_name,
            exDB.db_display_name,
            exDB.db_release,
            es.synonym,
            xref.info_type,
            xref.info_text,
            exDB.type,
            exDB.secondary_db_name,
            exDB.secondary_db_table,
            xref.description
    FROM    (xref, external_db exDB, dependent_xref dx, object_xref ox)
    LEFT JOIN external_synonym es ON
            es.xref_id = xref.xref_id
    WHERE   xref.external_db_id = exDB.external_db_id AND
            ox.xref_id = xref.xref_id AND
            ox.ensembl_object_type = \'$object_type\' AND
            ox.ensembl_id = ".$ensembl_object->dbID();

  if($constraint){
    $sql .= " AND $constraint";
  }
  else{
    die "NO constraint???\n";
  }
  my $sth = $self->prepare($sql) || die "Could not prepare $sql";
  return $self->_get_all_dm($sth);
}

sub _get_all_dm_sth {
  my ( $self, $constraint) = @_;

 my $sql = "SELECT xref.xref_id,
               xref.dbprimary_acc,
            xref.display_label,
            xref.version,
            exDB.priority,
            exDB.db_name,
            exDB.db_display_name,
            exDB.db_release,
            es.synonym,
            xref.info_type,
            xref.info_text,
            exDB.type,
            exDB.secondary_db_name,
            exDB.secondary_db_table,
            xref.description
    FROM    (xref, external_db exDB, dependent_xref dx)
    LEFT JOIN external_synonym es ON
            es.xref_id = xref.xref_id
    WHERE   xref.external_db_id = exDB.external_db_id ";

  if($constraint){
    $sql .= "AND $constraint";
  }
  else{
    die "NO constraint???\n";
  }

  my $sth = $self->prepare($sql) || die "Could not prepare $sql";

  return $self->_get_all_dm($sth);
}


sub _get_all_dm{

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

#  $sth->bind_param( 1, $dm_dbid, SQL_INTEGER );

#  print $sth."\n";
  $sth->execute() || die "Not able to execute statement handle";

  my @list =();
  my %seen;

  my $max_rows = 1000;
  while ( my $rowcache = $sth->fetchall_arrayref(undef, $max_rows) ) {
    while ( my $arrayref = shift( @{$rowcache} ) ) {
      my ( $dbID,                $dbprimaryId,
           $displayid,           $version,
           $dbname,              $db_display_name,
           $release,             $synonym,
           $info_type,           $info_text,
           $type,                $secondary_db_name,
           $secondary_db_table,  $description
      ) = @$arrayref;

      if ( !defined($seen{$dbID}) ) {
       my $exDB =
          Bio::EnsEMBL::DBEntry->new(
                           -adaptor             => $self,
                           -dbID                => $dbID,
                           -primary_id          => $dbprimaryId,
                           -display_id          => $displayid,
                           -version             => $version,
                           -release             => $release,
                           -dbname              => $dbname,
                           -priority            => $priority,
                           -db_display_name     => $db_display_name,
                           -info_type           => $info_type,
                           -info_text           => $info_text,
                           -type                => $type,
                           -secondary_db_name   => $secondary_db_name,
                           -secondary_db_table  => $secondary_db_table,
			   -description         => $description
          );

	if ($synonym) { $exDB->add_synonym($synonym) };
	$seen{$dbID} = 1;
	push @list, $exDB;
      }



    } ## end while ( my $arrayref = shift...
  } ## end while ( my $rowcache = $sth...

  $sth->finish();

  return \@list;

}


=head2 get_all_dependents

  Args[1]    : dbID of the DBentry to get the dependents of.
  Args[2]    : (optional) Bio::EnsEMBL::Gene, Transcript or Translation object
  Example    : my @dependents = @{ $dbe_adaptor->get_all_dependents(1234) };
  Description: Get a list of DBEntrys that are depenednet on the DBEntry.
               if an ensembl gene transcript or translation is given then only
               the ones on that object will be given
  Returntype : listref of DBEntrys. May be empty.
  Exceptions : none
  Caller     : DBEntry->get_all_dependnets
  Status     : UnStable

=cut

sub get_all_dependents {
  my ( $self, $dbid, $ensembl_object) = @_;
  
  if(defined($ensembl_object) and !($ensembl_object->isa("Bio::EnsEMBL::Feature") or $ensembl_object->isa("Bio::EnsEMBL::Translation"))){
    die ref($ensembl_object)." is not an Gene Transcript or Translation";
  }
  
  my $constraint = " dx.master_xref_id = $dbid AND  dx.dependent_xref_id = xref.xref_id";
  if(defined($ensembl_object)){
    return $self->_get_all_dm_loc_sth($constraint, $ensembl_object);
  }
  else{
    return $self->_get_all_dm_sth($constraint, $ensembl_object);
  }

}

=head2 get_all_masters

  Args[1]    : dbID of the DBentry to get the masters of.
  Args[2]    : (optional) Bio::EnsEMBL::Gene, Transcript or Translation object
  Example    : my @masters = @{ $dbe_adaptor->get_all_masters(1234) };
  Description: Get a list of DBEntrys that are the masters of the DBEntry.
               if an ensembl gene transcript or translation is given then only
               the ones on that object will be given.
  Returntype : listref of DBEntrys. May be empty.
  Exceptions : none
  Caller     : DBEntry->get_all_masters
  Status     : UnStable

=cut

sub get_all_masters {
  my ( $self, $dbid, $ensembl_object ) = @_;
  
  if(defined($ensembl_object) and !($ensembl_object->isa("Bio::EnsEMBL::Feature") or $ensembl_object->isa("Bio::EnsEMBL::Translation"))){
    die ref($ensembl_object)." is not an Gene Transcript or Translation";
  }
  
  my $constraint = "dx.dependent_xref_id = $dbid AND  dx.master_xref_id = xref.xref_id";

  if(defined($ensembl_object)){
    return $self->_get_all_dm_loc_sth($constraint, $ensembl_object);
  }
  else{
    return $self->_get_all_dm_sth($constraint, $ensembl_object);
  }
#  return $self->_get_all_dm($constraint, $ensembl_object);
}


=head fetch_all_by_name

  Arg [1]    : string $name - The name of the external reference.
               found in accession, display_label or synonym
  Arg [2]    : (optional) string $dbname  - The name of the database which 
               the provided name is for.

  Example    : my $xref = @{$dbea->fetch_all_by_name('BRAC2','HGNC')}[0];
               print $xref->description(), "\n" if($xref);
  Description: Retrieves list of DBEntrys (xrefs) via a name.
               The accesion is looked for first then the synonym and finally
               the display_label.
               NOTE $dbname this is optional but adding this speeds the
               process up if you know what you are looking for.

               NOTE:  In a multi-species database, this method will
               return all the entries matching the search criteria, not
               just the ones associated with the current species.
  Returntype : Bio::EnsEMBL::DBSQL::DBEntry
  Exceptions : thrown if arguments are incorrect
  Caller     : general, domainview
  Status     : Stable

=cut

sub fetch_all_by_name {
  my ( $self, $name, $dbname ) = @_;

  my $sql = (<<SQL);
  SELECT xref.xref_id, xref.dbprimary_acc, xref.display_label, xref.version,
         exDB.priority, exDB.db_name, exDB.db_display_name, exDB.db_release,
            es.synonym, xref.info_type, xref.info_text,
            exDB.type, exDB.secondary_db_name, exDB.secondary_db_table,
            xref.description
    FROM    (xref, external_db exDB)
    LEFT JOIN external_synonym es ON
            es.xref_id = xref.xref_id
    WHERE  (xref.dbprimary_acc = ? or xref.display_label = ?)
    AND    xref.external_db_id = exDB.external_db_id
SQL

  if(defined $dbname){
    $sql .= " AND    exDB.db_name = ?";
  }
  my $sth = $self->prepare($sql);
  $sth->bind_param( 1, $name, SQL_VARCHAR );
  $sth->bind_param( 2, $name, SQL_VARCHAR );
  if(defined $dbname){
    $sth->bind_param( 3 , $dbname,    SQL_VARCHAR );
  }
  $sth->execute();


  if ( !$sth->rows() && lc($dbname) eq 'interpro' ) {
  # This is a minor hack that means that results still come back even
  # when a mistake was made and no interpro accessions were loaded into
  # the xref table.  This has happened in the past and had the result of
  # breaking domainview

    $sth->finish();
    $sth = $self->prepare(
      "SELECT   NULL,
                i.interpro_ac,
                i.id,
                NULL,
                NULL,
                'Interpro',
                NULL,
                NULL
        FROM    interpro i
        WHERE   i.interpro_ac = ?" );

    $sth->bind_param( 1, $name, SQL_VARCHAR );
    $sth->execute();
  }

  my %exDB;
  my @exDBlist;
  my $max_rows = 1000;

  while ( my $rowcache = $sth->fetchall_arrayref( undef, $max_rows ) ) {
    while ( my $arrayref = shift( @{$rowcache} ) ) {
      my ( $dbID,                $dbprimaryId,
           $displayid,           $version,
           $priority,
           $dbname,              $db_display_name,
           $release,             $synonym,
           $info_type,           $info_text,
           $type,                $secondary_db_name,
           $secondary_db_table,  $description
      ) = @$arrayref;

      if ( !defined $exDB{$dbID} ) {
	my $entrie = 
          Bio::EnsEMBL::DBEntry->new(
                           -adaptor             => $self,
                           -dbID                => $dbID,
                           -primary_id          => $dbprimaryId,
                           -display_id          => $displayid,
                           -version             => $version,
                           -release             => $release,
                           -dbname              => $dbname,
                           -priority            => $priority,
                           -db_display_name     => $db_display_name,
                           -info_type           => $info_type,
                           -info_text           => $info_text,
                           -type                => $type,
                           -secondary_db_name   => $secondary_db_name,
                           -secondary_db_table  => $secondary_db_table,
			   -description         => $description
          );
	$exDB{$dbID} = $entrie;
	push @exDBlist, $entrie;
      }
      if ($synonym) { $exDB{$dbID}->add_synonym($synonym) }

    } ## end while ( my $arrayref = shift...
  } ## end while ( my $rowcache = $sth...

  $sth->finish();

  return \@exDBlist;
} ## end sub fetch_all_by_name



=head2 fetch_by_db_accession

  Arg [1]    : string $dbname - The name of the database which the provided
               accession is for.
  Arg [2]    : string $accession - The accesion of the external reference to
               retrieve.
  Example    : my $xref = $dbea->fetch_by_db_accession('Interpro','IPR003439');
               print $xref->description(), "\n" if($xref);
  Description: Retrieves a DBEntry (xref) via the name of the database
               it is from and its primary accession in that database.
               Undef is returned if the xref cannot be found in the
               database.
               NOTE:  In a multi-species database, this method will
               return all the entries matching the search criteria, not
               just the ones associated with the current species.
  Returntype : Bio::EnsEMBL::DBSQL::DBEntry
  Exceptions : thrown if arguments are incorrect
  Caller     : general, domainview

=cut

sub fetch_by_db_accession {
  my ( $self, $dbname, $accession ) = @_;
  my $sth = $self->prepare(
    "SELECT xref.xref_id,
            xref.dbprimary_acc,
            xref.display_label,
            xref.version,
            exDB.priority,
            exDB.db_name,
            exDB.db_display_name,
            exDB.db_release,
            es.synonym,
            xref.info_type,
            xref.info_text,
            exDB.type,
            exDB.secondary_db_name,
            xref.description
    FROM    (xref, external_db exDB)
    LEFT JOIN external_synonym es ON
            es.xref_id = xref.xref_id
    WHERE  xref.dbprimary_acc = ?
    AND    exDB.db_name = ?
    AND    xref.external_db_id = exDB.external_db_id" );
  $sth->bind_param( 1, $accession, SQL_VARCHAR );
  $sth->bind_param( 2, $dbname,    SQL_VARCHAR );
  if ( !$sth->rows() && lc($dbname) eq 'interpro' ) {
  # This is a minor hack that means that results still come back even
  # when a mistake was made and no interpro accessions were loaded into
  # the xref table.  This has happened in the past and had the result of
  # breaking domainview

    $sth->finish();
    $sth = $self->prepare(
      "SELECT   NULL,
                i.interpro_ac,
                i.id,
                NULL,
                NULL,
                'Interpro',
                NULL,
                NULL
        FROM    interpro i
        WHERE   i.interpro_ac = ?" );

    $sth->bind_param( 1, $accession, SQL_VARCHAR );
  my $max_rows = 1000;

  while ( my $rowcache = $sth->fetchall_arrayref( undef, $max_rows ) ) {
    while ( my $arrayref = shift( @{$rowcache} ) ) {
      my ( $dbID,                $dbprimaryId,
           $displayid,           $version,
           $dbname,              $db_display_name,
           $release,             $synonym,
           $info_type,           $info_text,
           $type,                $secondary_db_name,
      ) = @$arrayref;

      if ( !defined($exDB) ) {
        $exDB =
          Bio::EnsEMBL::DBEntry->new(
                           -adaptor             => $self,
                           -dbID                => $dbID,
                           -primary_id          => $dbprimaryId,
                           -display_id          => $displayid,
                           -version             => $version,
                           -release             => $release,
                           -dbname              => $dbname,
                           -priority            => $priority,
                           -db_display_name     => $db_display_name,
                           -info_type           => $info_type,
                           -info_text           => $info_text,
                           -type                => $type,
                           -secondary_db_name   => $secondary_db_name,
                           -secondary_db_table  => $secondary_db_table,
			   -description         => $description
      }

      if ($synonym) { $exDB->add_synonym($synonym) }

    } ## end while ( my $arrayref = shift...
  } ## end while ( my $rowcache = $sth...

  $sth->finish();
Graham McVicker's avatar
Graham McVicker committed
=head2 store

  Arg [1]    : Bio::EnsEMBL::DBEntry $dbEntry
               The DBEntry (xref) to be stored
  Arg [2]    : Int $ensID
               The dbID of an EnsEMBL object to associate with this external
               database entry
  Arg [3]    : string $ensType ('Transcript', 'Translation', 'Gene')
               The type of EnsEMBL object that this external database entry is
               being associated with.
  Arg [4]    : boolean $ignore_release
               If unset or zero, will require that the release string
               of the DBEntry object is identical to the release of the
               external database.  If set and non-zero, will ignore the
               release information.
  Example    : $dbea->store($db_entry, $transcript_id, 'Transcript');
  Description: Stores a reference to an external database (if it is not stored
               already) and associates an EnsEMBL object of a specified type
               with the external identifier.
  Returntype : int - the dbID of the newly created external refernce
  Exceptions : thrown when invalid dbID is passed to this method
  Caller     : scripts which load Xrefs and ObjectXrefs, etc. into Ensembl
Graham McVicker's avatar
Graham McVicker committed

=cut

  my ( $self, $dbEntry, $ensID, $ensType, $ignore_release ) = @_;
  #
  # backwards compatibility check:
  # check if $ensID is an object; if so, use $obj->dbID
  #
  my $ensembl_id;

  if ( defined($ensID) ) {
    if ( $ensID =~ /^\d+$/ ) {
      $ensembl_id = $ensID;
    } elsif (    ref($ensID) eq 'Bio::EnsEMBL::Gene'
              or ref($ensID) eq 'Bio::EnsEMBL::Transcript'
              or ref($ensID) eq 'Bio::EnsEMBL::Translation' 
              or ref($ensID) eq 'Bio::EnsEMBL::OperonTranscript'
              or ref($ensID) eq 'Bio::EnsEMBL::Operon' 
              )
    {
      warning(   "You should pass DBEntryAdaptor->store() "
               . "a dbID rather than an ensembl object "
               . "to store the xref on" );

      if ( defined( $ensID->dbID() ) ) {
        $ensembl_id = $ensID->dbID();
      } else {
        throw( sprintf( "%s %s doesn't have a dbID, can't store xref",
                        $ensType, $ensID->display_id() ) );
      }
      throw("Invalid dbID passed to DBEntryAdaptor->store()");
  
  
  
    # Ensure external_db contains a record of the intended xref source
    my $dbRef;
    $dbRef = $self->_check_external_db($dbEntry,$ignore_release);

    # Attempt to insert DBEntry
    my $xref_id = $self->_store_or_fetch_xref($dbEntry,$dbRef);
    $dbEntry->dbID($xref_id); #keeps DBEntry in sync with database
    ### Attempt to create an object->xref mapping
    if ($ensembl_id) {$self->_store_object_xref_mapping($ensembl_id,$dbEntry,$ensType)};
    
    return $xref_id;
}
    
sub _store_object_xref_mapping {
    my $self = shift;
    my $ensembl_id = shift;
    my $dbEntry = shift;
    my $ensembl_type = shift;
    
    if (not defined ($ensembl_type)) { warning("No Ensembl data type provided for new xref");}
    
    my $analysis_id;
    if ( $dbEntry->analysis() ) {
        $analysis_id = $self->db()->get_AnalysisAdaptor->store( $dbEntry->analysis() );
    } else {
        $analysis_id = 0; ## This used to be undef, but uniqueness in mysql requires a value
    
    my $sth = $self->prepare(qq(
        INSERT IGNORE INTO object_xref
          SET   xref_id = ?,
                ensembl_object_type = ?,
                ensembl_id = ?,
                linkage_annotation = ?,
                analysis_id = ? ) 
        );
    $sth->bind_param( 1, $dbEntry->dbID(),              SQL_INTEGER );
    $sth->bind_param( 2, $ensembl_type,                 SQL_VARCHAR );
    $sth->bind_param( 3, $ensembl_id,                   SQL_INTEGER );
    $sth->bind_param( 4, $dbEntry->linkage_annotation(),SQL_VARCHAR );
    $sth->bind_param( 5, $analysis_id,                  SQL_INTEGER );
    my $object_xref_id = $self->last_insert_id();
    
    $dbEntry->adaptor($self); # hand Adaptor to dbEntry for future use with OntologyXrefs
    
    if ($object_xref_id) {
        #no existing object_xref, therefore 
        if ( $dbEntry->isa('Bio::EnsEMBL::IdentityXref') ) {
             INSERT ignore INTO identity_xref
             SET object_xref_id = ?,
             xref_identity = ?,
             ensembl_identity = ?,
             xref_start = ?,
             xref_end   = ?,
             ensembl_start = ?,
             ensembl_end = ?,
        $sth->bind_param( 1, $object_xref_id,            SQL_INTEGER );
        $sth->bind_param( 2, $dbEntry->xref_identity,    SQL_INTEGER );
        $sth->bind_param( 3, $dbEntry->ensembl_identity, SQL_INTEGER );
        $sth->bind_param( 4, $dbEntry->xref_start,       SQL_INTEGER );
        $sth->bind_param( 5, $dbEntry->xref_end,         SQL_INTEGER );
        $sth->bind_param( 6, $dbEntry->ensembl_start,    SQL_INTEGER );
        $sth->bind_param( 7, $dbEntry->ensembl_end,      SQL_INTEGER );
        $sth->bind_param( 8,  $dbEntry->cigar_line,  SQL_LONGVARCHAR );
        $sth->bind_param( 9,  $dbEntry->score,            SQL_DOUBLE );
        $sth->bind_param( 10, $dbEntry->evalue,           SQL_DOUBLE );
      } elsif ( $dbEntry->isa('Bio::EnsEMBL::OntologyXref') ) {
                SET object_xref_id = ?,
                    linkage_type = ? " );
        foreach my $info ( @{ $dbEntry->get_all_linkage_info() } ) {
            my ( $linkage_type, $sourceXref ) = @{$info};
            my $sourceXid = undef;
            if ($sourceXref) {
              $sourceXref->is_stored( $self->dbc ) || $self->store($sourceXref);
              $sourceXid = $sourceXref->dbID;
            }
            $sth->bind_param( 1, $object_xref_id, SQL_INTEGER );
            $sth->bind_param( 2, $sourceXid, SQL_INTEGER );
            $sth->bind_param( 3, $linkage_type,  SQL_VARCHAR );
            $sth->execute();
        } #end foreach
      } #end elsif
    } # end if ($object_xref_id)
    return $object_xref_id;
}

=head2 _check_external_db 
  Arg [1]    : DBEntry object
  Arg [2]    : Ignore version flag
  Description: Looks for a record of the given external database
  Exceptions : Throws on missing external database entry
  Returntype : Int 

=cut

sub _check_external_db {
    my ($self,$db_entry,$ignore) = @_;
    my ($sql,@bound_params,$sql_helper,$db_name,$db_release);
    
    $db_name = $db_entry->dbname();
    $db_release = $db_entry->release();
    $sql_helper = $self->dbc->sql_helper;
    
    $sql = 'SELECT external_db_id FROM external_db WHERE db_name = ?';
    push @bound_params,$db_name;
    unless ($ignore) {
        if ($db_release) {
            $sql .= ' AND db_release = ?';
            push @bound_params,$db_release;
        } else {
            $sql .= ' AND db_release is NULL';
        }
    }
    
    my ($db_id) = @{ $sql_helper->execute_simple(-SQL => $sql, -PARAMS => \@bound_params) };
    
    if ($db_id) {
      return $db_id;
    }
    else {
      throw( sprintf( "external_db [%s] release [%s] does not exist",
                     $db_name, $db_release)
      );
    }
=head2 _store_or_fetch_xref

    Arg [1]    : DBEntry object
    Arg [2]    : Database accession for external database
    Description: Thread-safe method for adding xrefs, or otherwise returning
                 an xref ID for the inserted or retrieved xref. Also inserts
                 synonyms for that xref when entire new 
    Returns    : Int - the DB ID of the xref after insertion 
=cut
sub _store_or_fetch_xref {
    my $self = shift;
    my $dbEntry = shift;
    my $dbRef = shift;
    my $xref_id;
    
    my $sth = $self->prepare( "
       INSERT IGNORE INTO xref
       SET dbprimary_acc = ?,
           display_label = ?,
           version = ?,
           description = ?,
           external_db_id = ?,
           info_type = ?,
           info_text = ?");
    $sth->bind_param(1, $dbEntry->primary_id,SQL_VARCHAR);
    $sth->bind_param(2, $dbEntry->display_id,SQL_VARCHAR);
Andy Yates's avatar
Andy Yates committed
    $sth->bind_param(3, ($dbEntry->version || q{0}),SQL_VARCHAR);
    $sth->bind_param(4, $dbEntry->description,SQL_VARCHAR);
    $sth->bind_param(5, $dbRef,SQL_INTEGER);
    $sth->bind_param(6, ($dbEntry->info_type || 'NONE'), SQL_VARCHAR);
    $sth->bind_param(7, ($dbEntry->info_text || ''), SQL_VARCHAR);

    $sth->execute();
    $xref_id = $self->last_insert_id('xref_id',undef,'xref');
    $sth->finish();
    
    if ($xref_id) { #insert was successful, store supplementary synonyms
        # thread safety no longer an issue.
        my $synonym_check_sth = $self->prepare(
                  "SELECT xref_id, synonym
                   FROM external_synonym
                   WHERE xref_id = ?
                   AND synonym = ?");
    
        my $synonym_store_sth = $self->prepare(
            "INSERT ignore INTO external_synonym
             SET xref_id = ?, synonym = ?");
    
        my $synonyms = $dbEntry->get_all_synonyms();
        foreach my $syn ( @$synonyms ) {
            $synonym_check_sth->bind_param(1,$xref_id,SQL_INTEGER);
            $synonym_check_sth->bind_param(2,$syn,SQL_VARCHAR);
            $synonym_check_sth->execute();
            my ($dbSyn) = $synonym_check_sth->fetchrow_array();
            $synonym_store_sth->bind_param(1,$xref_id,SQL_INTEGER);
            $synonym_store_sth->bind_param(2,$syn,SQL_VARCHAR);
            $synonym_store_sth->execute() if(!$dbSyn);
        }
        $synonym_check_sth->finish();
        $synonym_store_sth->finish();
        
    } else { # xref_id already exists, retrieve it according to fields in the unique key
        my $info_type = $dbEntry->info_type() || 'NONE';
        my $info_text = $dbEntry->info_text() || q{};
        my $version = $dbEntry->version() || q{0};
        $sth->bind_param(1, $dbEntry->primary_id,SQL_VARCHAR);
        $sth->bind_param(2, $version, SQL_VARCHAR);
Andy Yates's avatar
Andy Yates committed
        $sth->bind_param(3, $dbRef, SQL_INTEGER);
        $sth->bind_param(4, $info_type, SQL_VARCHAR);
        $sth->bind_param(5, $info_text, SQL_VARCHAR);
        ($xref_id) = $sth->fetchrow_array();
        $sth->finish;
          my $msg = 'Cannot find an xref id for %s (version=%d) with external db id %d.';
          throw(sprintf($msg, $dbEntry->primary_id(), $version, $dbRef))
=head2 exists

  Arg [1]    : Bio::EnsEMBL::DBEntry $dbe
  Example    : if($dbID = $db_entry_adaptor->exists($dbe)) { do stuff; }
  Description: Returns the db id of this DBEntry if it exists in this database
               otherwise returns undef.  Exists is defined as an entry with
               the same external_db and display_id
  Returntype : int
  Exceptions : thrown on incorrect args
  Caller     : GeneAdaptor::store, TranscriptAdaptor::store
  my ($self, $dbe) = @_ ;

  unless($dbe && ref $dbe && $dbe->isa('Bio::EnsEMBL::DBEntry')) {
    throw("arg must be a Bio::EnsEMBL::DBEntry not [$dbe]");
  my $sth = $self->prepare('SELECT x.xref_id
                            FROM   xref x, external_db xdb
                            WHERE  x.external_db_id = xdb.external_db_id
                            AND    x.display_label = ?
                            AND    xdb.db_name = ?
                            AND    x.dbprimary_acc = ?');
  $sth->bind_param(1,$dbe->display_id,SQL_VARCHAR);
  $sth->bind_param(2,$dbe->dbname,SQL_VARCHAR);
  $sth->bind_param(3,$dbe->primary_id,SQL_VARCHAR);
Graham McVicker's avatar
Graham McVicker committed

               (The gene to retrieve DBEntries for)
  Arg [2]    : optional external database name. SQL wildcards are accepted
  Arg [3]    : optional external_db type. SQL wildcards are accepted
  Example    : @db_entries = @{$db_entry_adaptor->fetch_all_by_Gene($gene)};
  Description: This returns a list of DBEntries associated with this gene.
               Note that this method was changed in release 15.  Previously
               it set the DBLinks attribute of the gene passed in to contain
               all of the gene, transcript, and translation xrefs associated
  Returntype : listref of Bio::EnsEMBL::DBEntries; may be of type IdentityXref if
               there is mapping data, or OntologyXref if there is linkage data.
  Exceptions : thows if gene object not passed
Graham McVicker's avatar
Graham McVicker committed
  Caller     : Bio::EnsEMBL::Gene
Graham McVicker's avatar
Graham McVicker committed

=cut

  my ( $self, $gene, $ex_db_reg, $exdb_type ) = @_;
  if(!ref($gene) || !$gene->isa('Bio::EnsEMBL::Gene')) {
    throw("Bio::EnsEMBL::Gene argument expected.");
  }
  return $self->_fetch_by_object_type($gene->dbID(), 'Gene', $ex_db_reg, $exdb_type);
=head2 fetch_all_by_Operon

  Arg [1]    : Bio::EnsEMBL::Operon $operon
               (The operon to retrieve DBEntries for)
  Arg [2]    : optional external database name. SQL wildcards are accepted
  Arg [3]    : optional external_db type. SQL wildcards are accepted
  Example    : @db_entries = @{$db_entry_adaptor->fetch_all_by_Operon($operon)};
  Description: This returns a list of DBEntries associated with this operon.
  Returntype : listref of Bio::EnsEMBL::DBEntries; may be of type IdentityXref if
               there is mapping data, or OntologyXref if there is linkage data.
  Exceptions : thows if operon object not passed