my $self = shift;
# Now process the direct xrefs and add data to the object xrefs remember dependent xrefs.
my $object_xref_id;
my $dbi = $self->xref->dbc;
# First get the sths needed for the processing of the direct xrefs;
my $ins_ox_sql = (<<"IOS");
INSERT INTO object_xref (ensembl_id, xref_id, ensembl_object_type, linkage_type)
VALUES (?, ?, ?, ?)
IOS
my $ins_ox_sth = $dbi->prepare($ins_ox_sql);
# Direct xrefs can be considered to be 100% matching
my $ins_ix_sth = $self->get_ins_ix_sth($dbi);
my $stable_sql=(<<"SQL");
SELECT so.name, dx.general_xref_id, s.internal_id, dx.ensembl_stable_id , dx.linkage_xref
FROM source so, xref x, TYPE_direct_xref dx left join TYPE_stable_id s on s.stable_id = dx.ensembl_stable_id
WHERE x.xref_id = dx.general_xref_id and x.source_id = so.source_id
SQL
# We want to process the errors ourselves as for greater control
# If we get a error adding an object xref then it is already there
# This is not a problem. But we want to know how amny of these there were.
local $ins_ox_sth->{RaiseError} = 0; # want to see duplicates and not add de
local $ins_ox_sth->{PrintError} = 0;
my %err_count;
foreach my $table (qw(gene
transcript translation)){
my ($dbname, $xref_id, $internal_id, $stable_id, $linkage_type);
my $sql = $stable_sql;
$sql =~ s/TYPE/$table/g;
my $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$dbname, \$xref_id, \$internal_id, \$stable_id, \$linkage_type);
my $count =0;
my $duplicate_direct_count = 0;
my $duplicate_dependent_count = 0;
while($sth->fetch){
if(!defined($internal_id)){ # not found either it is an internal id already or stable_id no longer exists
if($stable_id =~ /^\d+$/){
$internal_id = $stable_id;
}
else{
if((!defined($err_count{$dbname})) or ($err_count{$dbname} < 10)){
print "Could not find stable id $stable_id in table to get the internal id hence ignoring!!! (for $dbname)\n" if($self->verbose);
}
$err_count{$dbname}++;
next;
}
}
$object_xref_id++;
$count++;
my @master_xref_ids;
if($internal_id == 0){
die "Problem could not find stable id $stable_id and got past the first check for $dbname\n";
}
$ins_ox_sth->execute($internal_id, $xref_id, $table, 'DIRECT');
$get_object_xref_id_sth->execute($internal_id, $xref_id, $table, 'DIRECT');
$object_xref_id = ($get_object_xref_id_sth->fetchrow_array())[0];
if($ins_ox_sth->err){
$duplicate_direct_count++;
next; #duplicate
}
else{
$ins_ix_sth->execute($object_xref_id);
push @master_xref_ids, $xref_id;
}
$self->process_dependents({master_xrefs => \@master_xref_ids,
dup_count => \$duplicate_dependent_count,
table => $table,
internal_id => $internal_id,
dbi => $dbi,
});
}
$sth->finish;
if($duplicate_direct_count or $duplicate_dependent_count){
print "duplicate entrys ignored for $duplicate_direct_count direct xrefs and $duplicate_dependent_count dependent xrefs\n" if($self->verbose);
}
}
foreach my $key ( keys %err_count){
print STDERR "*WARNING*: ".$err_count{$key}." direct xrefs for database ".$key." could not be added as their stable_ids could not be found\n";
}
my $sth = $dbi->prepare("insert into process_status (status, date) values('direct_xrefs_parsed',now())");
$sth->execute();
$sth->finish;
return;
}
Code:
