| 1459 |
# Get the database handle. |
# Get the database handle. |
| 1460 |
my $dbh = $fig->db_handle(); |
my $dbh = $fig->db_handle(); |
| 1461 |
# Ask for the synonyms. |
# Ask for the synonyms. |
| 1462 |
my $sth = $dbh->prepare_command("SELECT syn_id, maps_to FROM peg_synonyms ORDER BY syn_id"); |
my $sth = $dbh->prepare_command("SELECT maps_to, syn_id FROM peg_synonyms ORDER BY maps_to"); |
| 1463 |
my $result = $sth->execute(); |
my $result = $sth->execute(); |
| 1464 |
if (! defined($result)) { |
if (! defined($result)) { |
| 1465 |
Confess("Database error in Synonym load: " . $sth->errstr()); |
Confess("Database error in Synonym load: " . $sth->errstr()); |
| 1497 |
return $retVal; |
return $retVal; |
| 1498 |
} |
} |
| 1499 |
|
|
| 1500 |
|
=head3 LoadFamilyData |
| 1501 |
|
|
| 1502 |
|
C<< my $stats = $spl->LoadFamilyData(); >> |
| 1503 |
|
|
| 1504 |
|
Load the protein families into Sprout. |
| 1505 |
|
|
| 1506 |
|
The following relations are loaded by this method. |
| 1507 |
|
|
| 1508 |
|
Family |
| 1509 |
|
ContainsFeature |
| 1510 |
|
|
| 1511 |
|
The source information for these relations is taken from the C<families_for_protein>, |
| 1512 |
|
C<family_function>, and C<sz_family> methods of the B<FIG> object. |
| 1513 |
|
|
| 1514 |
|
=over 4 |
| 1515 |
|
|
| 1516 |
|
=item RETURNS |
| 1517 |
|
|
| 1518 |
|
Returns a statistics object for the loads. |
| 1519 |
|
|
| 1520 |
|
=back |
| 1521 |
|
|
| 1522 |
|
=cut |
| 1523 |
|
#: Return Type $%; |
| 1524 |
|
sub LoadFamilyData { |
| 1525 |
|
# Get this object instance. |
| 1526 |
|
my ($self) = @_; |
| 1527 |
|
# Get the FIG object. |
| 1528 |
|
my $fig = $self->{fig}; |
| 1529 |
|
# Get the genome hash. |
| 1530 |
|
my $genomeHash = $self->{genomes}; |
| 1531 |
|
# Create load objects for the tables we're loading. |
| 1532 |
|
my $loadFamily = $self->_TableLoader('Family'); |
| 1533 |
|
my $loadContainsFeature = $self->_TableLoader('ContainsFeature'); |
| 1534 |
|
if ($self->{options}->{loadOnly}) { |
| 1535 |
|
Trace("Loading from existing files.") if T(2); |
| 1536 |
|
} else { |
| 1537 |
|
Trace("Generating family data.") if T(2); |
| 1538 |
|
# Create a hash for the family IDs. |
| 1539 |
|
my %familyHash = (); |
| 1540 |
|
# Loop through the genomes. |
| 1541 |
|
for my $genomeID (sort keys %{$genomeHash}) { |
| 1542 |
|
Trace("Processing features for $genomeID.") if T(2); |
| 1543 |
|
# Loop through this genome's PEGs. |
| 1544 |
|
for my $fid ($fig->all_features($genomeID, "peg")) { |
| 1545 |
|
$loadContainsFeature->Add("features", 1); |
| 1546 |
|
# Get this feature's families. |
| 1547 |
|
my @families = $fig->families_for_protein($fid); |
| 1548 |
|
# Loop through the families, connecting them to the feature. |
| 1549 |
|
for my $family (@families) { |
| 1550 |
|
$loadContainsFeature->Put($family, $fid); |
| 1551 |
|
# If this is a new family, create a record for it. |
| 1552 |
|
if (! exists $familyHash{$family}) { |
| 1553 |
|
$familyHash{$family} = 1; |
| 1554 |
|
$loadFamily->Add("families", 1); |
| 1555 |
|
my $size = $fig->sz_family($family); |
| 1556 |
|
my $func = $fig->family_function($family); |
| 1557 |
|
$loadFamily->Put($family, $size, $func); |
| 1558 |
|
} |
| 1559 |
|
} |
| 1560 |
|
} |
| 1561 |
|
} |
| 1562 |
|
} |
| 1563 |
|
# Finish the load. |
| 1564 |
|
my $retVal = $self->_FinishAll(); |
| 1565 |
|
return $retVal; |
| 1566 |
|
} |
| 1567 |
|
|
| 1568 |
=head2 Internal Utility Methods |
=head2 Internal Utility Methods |
| 1569 |
|
|
Parent Directory
| 
Revision Log
| 
Patch