Diff of /Sprout/SproutLoad.pm

-revision 1.26, Mon Jan 30 21:57:02 2006 UTC
+revision 1.94, Tue Sep 16 18:42:40 2008 UTC
 Line 7
      use PageBuilder;
      use ERDBLoad;
      use FIG;
+     use FIGRules;
      use Sprout;
      use Stats;
      use BasicLocation;
      use HTML;
+     use AliasAnalysis;
+     use BioWords;
  =head1 Sprout Load Methods
-Line 30
+Line 33
      $stats->Accumulate($spl->LoadFeatureData());
      print $stats->Show();
- This module makes use of the internal Sprout property C<_erdb>.
  It is worth noting that the FIG object does not need to be a real one. Any object
  that implements the FIG methods for data retrieval could be used. So, for example,
  this object could be used to copy data from one Sprout database to another, or
-Line 52
+Line 53
  =head3 new
- C<< my $spl = SproutLoad->new($sprout, $fig, $genomeFile, $subsysFile, $options); >>
+     my $spl = SproutLoad->new($sprout, $fig, $genomeFile, $subsysFile, $options);
  Construct a new Sprout Loader object, specifying the two participating databases and
  the name of the files containing the list of genomes and subsystems to use.
-Line 80
+Line 81
  =item subsysFile
  Either the name of the file containing the list of trusted subsystems or a reference
- to a list of subsystem names. If nothing is specified, all known subsystems will be
+ to a list of subsystem names. If nothing is specified, all NMPDR subsystems will be
- considered trusted. Only subsystem data related to the trusted subsystems is loaded.
+ considered trusted. (A subsystem is considered NMPDR if it has a file named C<NMPDR>
+ in its data directory.) Only subsystem data related to the NMPDR subsystems is loaded.
  =item options
-Line 94
+Line 96
  sub new {
      # Get the parameters.
      my ($class, $sprout, $fig, $genomeFile, $subsysFile, $options) = @_;
-     # Load the list of genomes into a hash.
+     # Create the genome hash.
-     my %genomes;
+     my %genomes = ();
+     # We only need it if load-only is NOT specified.
+     if (! $options->{loadOnly}) {
      if (! defined($genomeFile) || $genomeFile eq '') {
          # Here we want all the complete genomes and an access code of 1.
          my @genomeList = $fig->genomes(1);
          %genomes = map { $_ => 1 } @genomeList;
+             Trace(scalar(keys %genomes) . " genomes found.") if T(3);
      } else {
          my $type = ref $genomeFile;
          Trace("Genome file parameter type is \"$type\".") if T(3);
-Line 119
+Line 124
                  # an omitted access code can be defaulted to 1.
                  for my $genomeLine (@genomeList) {
                      my ($genomeID, $accessCode) = split("\t", $genomeLine);
-                     if (undef $accessCode) {
+                         if (! defined($accessCode)) {
                          $accessCode = 1;
                      }
                      $genomes{$genomeID} = $accessCode;
-Line 129
+Line 134
              Confess("Invalid genome parameter ($type) in SproutLoad constructor.");
          }
      }
+     }
      # Load the list of trusted subsystems.
      my %subsystems = ();
+     # We only need it if load-only is NOT specified.
+     if (! $options->{loadOnly}) {
      if (! defined $subsysFile || $subsysFile eq '') {
-         # Here we want all the subsystems.
+             # Here we want all the usable subsystems. First we get the whole list.
-         %subsystems = map { $_ => 1 } $fig->all_subsystems();
+             my @subs = $fig->all_subsystems();
+             # Loop through, checking for the NMPDR file.
+             for my $sub (@subs) {
+                 if ($fig->nmpdr_subsystem($sub)) {
+                     $subsystems{$sub} = 1;
+                 }
+             }
      } else {
          my $type = ref $subsysFile;
          if ($type eq 'ARRAY') {
-Line 153
+Line 167
              Confess("Invalid subsystem parameter in SproutLoad constructor.");
          }
      }
+         # Go through the subsys hash again, creating the keyword list for each subsystem.
+         for my $subsystem (keys %subsystems) {
+             my $name = $subsystem;
+             $name =~ s/_/ /g;
+             $subsystems{$subsystem} = $name;
+         }
+     }
+     # Get the list of NMPDR-oriented attribute keys.
+     my @propKeys = $fig->get_group_keys("NMPDR");
      # Get the data directory from the Sprout object.
      my ($directory) = $sprout->LoadInfo();
      # Create the Sprout load object.
-Line 162
+Line 185
                    subsystems => \%subsystems,
                    sprout => $sprout,
                    loadDirectory => $directory,
-                   erdb => $sprout->{_erdb},
+                   erdb => $sprout,
                    loaders => [],
-                   options => $options
+                   options => $options,
+                   propKeys => \@propKeys,
                   };
      # Bless and return it.
      bless $retVal, $class;
-Line 173
+Line 197
  =head3 LoadOnly
- C<< my $flag = $spl->LoadOnly; >>
+     my $flag = $spl->LoadOnly;
  Return TRUE if we are in load-only mode, else FALSE.
-Line 184
+Line 208
      return $self->{options}->{loadOnly};
  }
- =head3 PrimaryOnly
- C<< my $flag = $spl->PrimaryOnly; >>
- Return TRUE if only the main entity is to be loaded, else FALSE.
- =cut
- sub PrimaryOnly {
-     my ($self) = @_;
-     return $self->{options}->{primaryOnly};
- }
  =head3 LoadGenomeData
- C<< my $stats = $spl->LoadGenomeData(); >>
+     my $stats = $spl->LoadGenomeData();
  Load the Genome, Contig, and Sequence data from FIG into Sprout.
-Line 236
+Line 248
      my $genomeCount = (keys %{$genomeHash});
      # Create load objects for each of the tables we're loading.
      my $loadGenome = $self->_TableLoader('Genome');
-     my $loadHasContig = $self->_TableLoader('HasContig', $self->PrimaryOnly);
+     my $loadHasContig = $self->_TableLoader('HasContig');
-     my $loadContig = $self->_TableLoader('Contig', $self->PrimaryOnly);
+     my $loadContig = $self->_TableLoader('Contig');
-     my $loadIsMadeUpOf = $self->_TableLoader('IsMadeUpOf', $self->PrimaryOnly);
+     my $loadIsMadeUpOf = $self->_TableLoader('IsMadeUpOf');
-     my $loadSequence = $self->_TableLoader('Sequence', $self->PrimaryOnly);
+     my $loadSequence = $self->_TableLoader('Sequence');
      if ($self->{options}->{loadOnly}) {
          Trace("Loading from existing files.") if T(2);
      } else {
          Trace("Generating genome data.") if T(2);
+         # Get the full info for the FIG genomes.
+         my %genomeInfo = map { $_->[0] => { gname => $_->[1], szdna => $_->[2], maindomain => $_->[3],
+                                             pegs => $_->[4], rnas => $_->[5], complete => $_->[6] } } @{$fig->genome_info()};
          # Now we loop through the genomes, generating the data for each one.
          for my $genomeID (sort keys %{$genomeHash}) {
              Trace("Generating data for genome $genomeID.") if T(3);
              $loadGenome->Add("genomeIn");
              # The access code comes in via the genome hash.
              my $accessCode = $genomeHash->{$genomeID};
-             # Get the genus, species, and strain from the scientific name. Note that we append
+             # Get the genus, species, and strain from the scientific name.
-             # the genome ID to the strain. In some cases this is the totality of the strain name.
              my ($genus, $species, @extraData) = split / /, $self->{fig}->genus_species($genomeID);
-             my $extra = join " ", @extraData, "[$genomeID]";
+             my $extra = join " ", @extraData;
              # Get the full taxonomy.
              my $taxonomy = $fig->taxonomy_of($genomeID);
+             # Get the version. If no version is specified, we default to the genome ID by itself.
+             my $version = $fig->genome_version($genomeID);
+             if (! defined($version)) {
+                 $version = $genomeID;
+             }
+             # Get the DNA size.
+             my $dnaSize = $fig->genome_szdna($genomeID);
+             # Open the NMPDR group file for this genome.
+             my $group;
+             if (open(TMP, "<$FIG_Config::organisms/$genomeID/NMPDR") &&
+                 defined($group = <TMP>)) {
+                 # Clean the line ending.
+                 chomp $group;
+             } else {
+                 # No group, so use the default.
+                 $group = $FIG_Config::otherGroup;
+             }
+             close TMP;
+             # Get the contigs.
+             my @contigs = $fig->all_contigs($genomeID);
+             # Get this genome's info array.
+             my $info = $genomeInfo{$genomeID};
              # Output the genome record.
-             $loadGenome->Put($genomeID, $accessCode, $fig->is_complete($genomeID), $genus,
+             $loadGenome->Put($genomeID, $accessCode, $info->{complete}, scalar(@contigs),
-                              $species, $extra, $taxonomy);
+                              $dnaSize, $genus, $info->{pegs}, $group, $info->{rnas}, $species, $extra, $version, $taxonomy);
              # Now we loop through each of the genome's contigs.
-             my @contigs = $fig->all_contigs($genomeID);
              for my $contigID (@contigs) {
                  Trace("Processing contig $contigID for $genomeID.") if T(4);
                  $loadContig->Add("contigIn");
-Line 296
+Line 331
      return $retVal;
  }
- =head3 LoadCouplingData
- C<< my $stats = $spl->LoadCouplingData(); >>
- Load the coupling and evidence data from FIG into Sprout.
- The coupling data specifies which genome features are functionally coupled. The
- evidence data explains why the coupling is functional.
- The following relations are loaded by this method.
-     Coupling
-     IsEvidencedBy
-     PCH
-     ParticipatesInCoupling
-     UsesAsEvidence
- =over 4
- =item RETURNS
- Returns a statistics object for the loads.
- =back
- =cut
- #: Return Type $%;
- sub LoadCouplingData {
-     # Get this object instance.
-     my ($self) = @_;
-     # Get the FIG object.
-     my $fig = $self->{fig};
-     # Get the genome hash.
-     my $genomeFilter = $self->{genomes};
-     my $genomeCount = (keys %{$genomeFilter});
-     my $featureCount = $genomeCount * 4000;
-     # Start the loads.
-     my $loadCoupling = $self->_TableLoader('Coupling');
-     my $loadIsEvidencedBy = $self->_TableLoader('IsEvidencedBy', $self->PrimaryOnly);
-     my $loadPCH = $self->_TableLoader('PCH', $self->PrimaryOnly);
-     my $loadParticipatesInCoupling = $self->_TableLoader('ParticipatesInCoupling', $self->PrimaryOnly);
-     my $loadUsesAsEvidence = $self->_TableLoader('UsesAsEvidence', $self->PrimaryOnly);
-     if ($self->{options}->{loadOnly}) {
-         Trace("Loading from existing files.") if T(2);
-     } else {
-         Trace("Generating coupling data.") if T(2);
-         # Loop through the genomes found.
-         for my $genome (sort keys %{$genomeFilter}) {
-             Trace("Generating coupling data for $genome.") if T(3);
-             $loadCoupling->Add("genomeIn");
-             # Create a hash table for holding coupled pairs. We use this to prevent
-             # duplicates. For example, if A is coupled to B, we don't want to also
-             # assert that B is coupled to A, because we already know it. Fortunately,
-             # all couplings occur within a genome, so we can keep the hash table
-             # size reasonably small.
-             my %dupHash = ();
-             # Get all of the genome's PEGs.
-             my @pegs = $fig->pegs_of($genome);
-             # Loop through the PEGs.
-             for my $peg1 (@pegs) {
-                 $loadCoupling->Add("pegIn");
-                 Trace("Processing PEG $peg1 for $genome.") if T(4);
-                 # Get a list of the coupled PEGs.
-                 my @couplings = $fig->coupled_to($peg1);
-                 # For each coupled PEG, we need to verify that a coupling already
-                 # exists. If not, we have to create one.
-                 for my $coupleData (@couplings) {
-                     my ($peg2, $score) = @{$coupleData};
-                     # Compute the coupling ID.
-                     my $coupleID = Sprout::CouplingID($peg1, $peg2);
-                     if (! exists $dupHash{$coupleID}) {
-                         $loadCoupling->Add("couplingIn");
-                         # Here we have a new coupling to store in the load files.
-                         Trace("Storing coupling ($coupleID) with score $score.") if T(4);
-                         # Ensure we don't do this again.
-                         $dupHash{$coupleID} = $score;
-                         # Write the coupling record.
-                         $loadCoupling->Put($coupleID, $score);
-                         # Connect it to the coupled PEGs.
-                         $loadParticipatesInCoupling->Put($peg1, $coupleID, 1);
-                         $loadParticipatesInCoupling->Put($peg2, $coupleID, 2);
-                         # Get the evidence for this coupling.
-                         my @evidence = $fig->coupling_evidence($peg1, $peg2);
-                         # Organize the evidence into a hash table.
-                         my %evidenceMap = ();
-                         # Process each evidence item.
-                         for my $evidenceData (@evidence) {
-                             $loadPCH->Add("evidenceIn");
-                             my ($peg3, $peg4, $usage) = @{$evidenceData};
-                             # Only proceed if the evidence is from a Sprout
-                             # genome.
-                             if ($genomeFilter->{$fig->genome_of($peg3)}) {
-                                 $loadUsesAsEvidence->Add("evidenceChosen");
-                                 my $evidenceKey = "$coupleID $peg3 $peg4";
-                                 # We store this evidence in the hash if the usage
-                                 # is nonzero or no prior evidence has been found. This
-                                 # insures that if there is duplicate evidence, we
-                                 # at least keep the meaningful ones. Only evidence in
-                                 # the hash makes it to the output.
-                                 if ($usage || ! exists $evidenceMap{$evidenceKey}) {
-                                     $evidenceMap{$evidenceKey} = $evidenceData;
-                                 }
-                             }
-                         }
-                         for my $evidenceID (keys %evidenceMap) {
-                             # Create the evidence record.
-                             my ($peg3, $peg4, $usage) = @{$evidenceMap{$evidenceID}};
-                             $loadPCH->Put($evidenceID, $usage);
-                             # Connect it to the coupling.
-                             $loadIsEvidencedBy->Put($coupleID, $evidenceID);
-                             # Connect it to the features.
-                             $loadUsesAsEvidence->Put($evidenceID, $peg3, 1);
-                             $loadUsesAsEvidence->Put($evidenceID, $peg4, 2);
-                         }
-                     }
-                 }
-             }
-         }
-     }
-     # All done. Finish the load.
-     my $retVal = $self->_FinishAll();
-     return $retVal;
- }
  =head3 LoadFeatureData
- C<< my $stats = $spl->LoadFeatureData(); >>
+     my $stats = $spl->LoadFeatureData();
  Load the feature data from FIG into Sprout.
-Line 432
+Line 343
      Feature
      FeatureAlias
+     IsAliasOf
      FeatureLink
      FeatureTranslation
      FeatureUpstream
      IsLocatedIn
+     HasFeature
+     HasRoleInSubsystem
+     FeatureEssential
+     FeatureVirulent
+     FeatureIEDB
+     CDD
+     IsPresentOnProteinOf
+     CellLocation
+     IsPossiblePlaceFor
+     ExternalDatabase
+     IsAlsoFoundIn
+     Keyword
  =over 4
-Line 450
+Line 374
  sub LoadFeatureData {
      # Get this object instance.
      my ($self) = @_;
-     # Get the FIG object.
+     # Get the FIG and Sprout objects.
      my $fig = $self->{fig};
+     my $sprout = $self->{sprout};
      # Get the table of genome IDs.
      my $genomeHash = $self->{genomes};
      # Create load objects for each of the tables we're loading.
      my $loadFeature = $self->_TableLoader('Feature');
-     my $loadIsLocatedIn = $self->_TableLoader('IsLocatedIn', $self->PrimaryOnly);
+     my $loadIsLocatedIn = $self->_TableLoader('IsLocatedIn');
      my $loadFeatureAlias = $self->_TableLoader('FeatureAlias');
+     my $loadIsAliasOf = $self->_TableLoader('IsAliasOf');
      my $loadFeatureLink = $self->_TableLoader('FeatureLink');
      my $loadFeatureTranslation = $self->_TableLoader('FeatureTranslation');
      my $loadFeatureUpstream = $self->_TableLoader('FeatureUpstream');
+     my $loadHasFeature = $self->_TableLoader('HasFeature');
+     my $loadHasRoleInSubsystem = $self->_TableLoader('HasRoleInSubsystem');
+     my $loadFeatureEssential = $self->_TableLoader('FeatureEssential');
+     my $loadFeatureVirulent = $self->_TableLoader('FeatureVirulent');
+     my $loadFeatureIEDB = $self->_TableLoader('FeatureIEDB');
+     my $loadCDD = $self->_TableLoader('CDD');
+     my $loadIsPresentOnProteinOf = $self->_TableLoader('IsPresentOnProteinOf');
+     my $loadCellLocation = $self->_TableLoader('CellLocation');
+     my $loadIsPossiblePlaceFor = $self->_TableLoader('IsPossiblePlaceFor');
+     my $loadIsAlsoFoundIn = $self->_TableLoader('IsAlsoFoundIn');
+     my $loadExternalDatabase = $self->_TableLoader('ExternalDatabase');
+     my $loadKeyword = $self->_TableLoader('Keyword');
+     # Get the subsystem hash.
+     my $subHash = $self->{subsystems};
+     # Get the property keys.
+     my $propKeys = $self->{propKeys};
+     # Create a hashes to hold CDD, Cell Location (PSORT), External Database, and alias values.
+     my %CDD = ();
+     my %alias = ();
+     my %cellLocation = ();
+     my %xdb = ();
+     # Create the bio-words object.
+     my $biowords = BioWords->new(exceptions => "$FIG_Config::sproutData/Exceptions.txt",
+                                  stops => "$FIG_Config::sproutData/StopWords.txt",
+                                  cache => 0);
+     # One of the things we have to do here is build the keyword table, and the keyword
+     # table needs to contain the originating text and feature count for each stem. Unfortunately,
+     # the number of distinct keywords is so large it causes PERL to hang if we try to
+     # keep them in memory. As a result, we need to track them using disk files.
+     # Our approach will be to use two sequential files. One will contain stems and phonexes.
+     # Each time a stem occurs in a feature, a record will be written to that file. The stem
+     # file can then be sorted and collated to determine the number of features for each
+     # stem. A separate file will contain keywords and stems. This last file
+     # will be subjected to a sort unique on stem/keyword. The file is then merged
+     # with the stem file to create the keyword table relation (keyword, stem, phonex, count).
+     my $stemFileName = "$FIG_Config::temp/stems$$.tbl";
+     my $keyFileName = "$FIG_Config::temp/keys$$.tbl";
+     my $stemh = Open(undef, "| sort -T\"$FIG_Config::temp\" -t\"\t\" -k1,1 >$stemFileName");
+     my $keyh = Open(undef, "| sort -T\"$FIG_Config::temp\" -t\"\t\" -u -k1,1 -k2,2 >$keyFileName");
      # Get the maximum sequence size. We need this later for splitting up the
      # locations.
      my $chunkSize = $self->{sprout}->MaxSegment();
-Line 469
+Line 434
      } else {
          Trace("Generating feature data.") if T(2);
          # Now we loop through the genomes, generating the data for each one.
-         for my $genomeID (sort keys %{$genomeHash}) {
+         my @allGenomes = sort keys %{$genomeHash};
+         Trace(scalar(@allGenomes) . " genomes found in list.") if T(3);
+         for my $genomeID (@allGenomes) {
              Trace("Loading features for genome $genomeID.") if T(3);
              $loadFeature->Add("genomeIn");
              # Get the feature list for this genome.
-             my $features = $fig->all_features_detailed($genomeID);
+             my $features = $fig->all_features_detailed_fast($genomeID);
+             # Sort and count the list.
+             my @featureTuples = sort { $a->[0] cmp $b->[0] } @{$features};
+             my $count = scalar @featureTuples;
+             my @fids = map { $_->[0] } @featureTuples;
+             Trace("$count features found for genome $genomeID.") if T(3);
+             # Get the attributes for this genome and put them in a hash by feature ID.
+             my $attributes = GetGenomeAttributes($fig, $genomeID, \@fids, $propKeys);
+             Trace("Looping through features for $genomeID.") if T(3);
+             # Set up for our duplicate-feature check.
+             my $oldFeatureID = "";
              # Loop through the features.
-             for my $featureData (@{$features}) {
+             for my $featureTuple (@featureTuples) {
-                 $loadFeature->Add("featureIn");
                  # Split the tuple.
-                 my ($featureID, $locations, undef, $type) = @{$featureData};
+                 my ($featureID, $locations, undef, $type, $minloc, $maxloc, $assignment, $user, $quality) = @{$featureTuple};
-                 # Create the feature record.
+                 # Check for duplicates.
-                 $loadFeature->Put($featureID, 1, $type);
+                 if ($featureID eq $oldFeatureID) {
+                     Trace("Duplicate feature $featureID found.") if T(1);
+                 } else {
+                     $oldFeatureID = $featureID;
+                     # Count this feature.
+                     $loadFeature->Add("featureIn");
+                     # Fix the quality. It is almost always a space, but some odd stuff might sneak through, and the
+                     # Sprout database requires a single character.
+                     if (! defined($quality) || $quality eq "") {
+                         $quality = " ";
+                     }
+                     # Begin building the keywords. We start with the genome ID, the
+                     # feature ID, the taxonomy, and the organism name.
+                     my @keywords = ($genomeID, $featureID, $fig->genus_species($genomeID),
+                                     $fig->taxonomy_of($genomeID));
                  # Create the aliases.
                  for my $alias ($fig->feature_aliases($featureID)) {
-                     $loadFeatureAlias->Put($featureID, $alias);
+                         #Connect this alias to this feature.
+                         $loadIsAliasOf->Put($alias, $featureID);
+                         push @keywords, $alias;
+                         # If this is a locus tag, also add its natural form as a keyword.
+                         my $naturalName = AliasAnalysis::Type(LocusTag => $alias);
+                         if ($naturalName) {
+                             push @keywords, $naturalName;
+                         }
+                         # If this is the first time for the specified alias, create its
+                         # alias record.
+                         if (! exists $alias{$alias}) {
+                             $loadFeatureAlias->Put($alias);
+                             $alias{$alias} = 1;
+                         }
+                     }
+                     # Add the corresponding IDs. We ask for 2-tuples of the form (id, database).
+                     my @corresponders = $fig->get_corresponding_ids($featureID, 1);
+                     for my $tuple (@corresponders) {
+                         my ($id, $xdb) = @{$tuple};
+                         # Ignore SEED: that's us.
+                         if ($xdb ne 'SEED') {
+                             # Connect this ID to the feature.
+                             $loadIsAlsoFoundIn->Put($featureID, $xdb, $id);
+                             # Add it as a keyword.
+                             push @keywords, $id;
+                             # If this is a new database, create a record for it.
+                             if (! exists $xdb{$xdb}) {
+                                 $xdb{$xdb} = 1;
+                                 $loadExternalDatabase->Put($xdb);
+                             }
                  }
+                     }
+                     Trace("Assignment for $featureID is: $assignment") if T(4);
+                     # Break the assignment into words and shove it onto the
+                     # keyword list.
+                     push @keywords, split(/\s+/, $assignment);
+                     # Link this feature to the parent genome.
+                     $loadHasFeature->Put($genomeID, $featureID, $type);
                  # Get the links.
                  my @links = $fig->fid_links($featureID);
                  for my $link (@links) {
-Line 503
+Line 529
                          $loadFeatureUpstream->Put($featureID, $upstream);
                      }
                  }
+                     # Now we need to find the subsystems this feature participates in.
+                     # We also add the subsystems to the keyword list. Before we do that,
+                     # we must convert underscores to spaces.
+                     my @subsystems = $fig->peg_to_subsystems($featureID);
+                     for my $subsystem (@subsystems) {
+                         # Only proceed if we like this subsystem.
+                         if (exists $subHash->{$subsystem}) {
+                             # Store the has-role link.
+                             $loadHasRoleInSubsystem->Put($featureID, $subsystem, $genomeID, $type);
+                             # Save the subsystem's keyword data.
+                             my $subKeywords = $subHash->{$subsystem};
+                             push @keywords, split /\s+/, $subKeywords;
+                             # Now we need to get this feature's role in the subsystem.
+                             my $subObject = $fig->get_subsystem($subsystem);
+                             my @roleColumns = $subObject->get_peg_roles($featureID);
+                             my @allRoles = $subObject->get_roles();
+                             for my $col (@roleColumns) {
+                                 my $role = $allRoles[$col];
+                                 push @keywords, split /\s+/, $role;
+                                 push @keywords, $subObject->get_role_abbr($col);
+                             }
+                         }
+                     }
+                     # There are three special attributes computed from property
+                     # data that we build next. If the special attribute is non-empty,
+                     # its name will be added to the keyword list. First, we get all
+                     # the attributes for this feature. They will come back as
+                     # 4-tuples: [peg, name, value, URL]. We use a 3-tuple instead:
+                     # [name, value, value with URL]. (We don't need the PEG, since
+                     # we already know it.)
+                     my @attributes = map { [$_->[1], $_->[2], Tracer::CombineURL($_->[2], $_->[3])] }
+                                          @{$attributes->{$featureID}};
+                     # Now we process each of the special attributes.
+                     if (SpecialAttribute($featureID, \@attributes,
+, [0,2], '^(essential|potential_essential)$',
+                                          $loadFeatureEssential)) {
+                         push @keywords, 'essential';
+                         $loadFeature->Add('essential');
+                     }
+                     if (SpecialAttribute($featureID, \@attributes,
+, [2], '^virulen',
+                                          $loadFeatureVirulent)) {
+                         push @keywords, 'virulent';
+                         $loadFeature->Add('virulent');
+                     }
+                     if (SpecialAttribute($featureID, \@attributes,
+, [0,2], '^iedb_',
+                                          $loadFeatureIEDB)) {
+                         push @keywords, 'iedb';
+                         $loadFeature->Add('iedb');
+                     }
+                     # Now we have some other attributes we need to process. To get
+                     # through them, we convert the attribute list for this feature
+                     # into a two-layer hash: key => subkey => value.
+                     my %attributeHash = ();
+                     for my $attrRow (@{$attributes->{$featureID}}) {
+                         my (undef, $key, @values) = @{$attrRow};
+                         my ($realKey, $subKey);
+                         if ($key =~ /^([^:]+)::(.+)/) {
+                             ($realKey, $subKey) = ($1, $2);
+                         } else {
+                             ($realKey, $subKey) = ($key, "");
+                         }
+                         if (exists $attributeHash{$1}) {
+                             $attributeHash{$1}->{$2} = \@values;
+                         } else {
+                             $attributeHash{$1} = {$2 => \@values};
+                         }
+                     }
+                     # First we handle CDD. This is a bit complicated, because
+                     # there are multiple CDDs per protein.
+                     if (exists $attributeHash{CDD}) {
+                         # Get the hash of CDD IDs to scores for this feature. We
+                         # already know it exists because of the above IF.
+                         my $cddHash = $attributeHash{CDD};
+                         my @cddData = sort keys %{$cddHash};
+                         for my $cdd (@cddData) {
+                             # Extract the score for this CDD and decode it.
+                             my ($codeScore) = split(/\s*[,;]\s*/, $cddHash->{$cdd}->[0]);
+                             my $realScore = FIGRules::DecodeScore($codeScore);
+                             # We can't afford to crash because of a bad attribute
+                             # value, hence the IF below.
+                             if (! defined($realScore)) {
+                                 # Bad score, so count it.
+                                 $loadFeature->Add('badCDDscore');
+                                 Trace("CDD score \"$codeScore\" for feature $featureID invalid.") if T(3);
+                             } else {
+                                 # Create the connection.
+                                 $loadIsPresentOnProteinOf->Put($cdd, $featureID, $realScore);
+                                 # If this CDD does not yet exist, create its record.
+                                 if (! exists $CDD{$cdd}) {
+                                     $CDD{$cdd} = 1;
+                                     $loadCDD->Put($cdd);
+                                 }
+                             }
+                         }
+                     }
+                     # Next we do PSORT cell locations. here the confidence value
+                     # could have the value "unknown", which we translate to -1.
+                     if (exists $attributeHash{PSORT}) {
+                         # This will be a hash of cell locations to confidence
+                         # factors.
+                         my $psortHash = $attributeHash{PSORT};
+                         for my $psort (keys %{$psortHash}) {
+                             # Get the confidence, and convert it to a number if necessary.
+                             my $confidence = $psortHash->{$psort};
+                             if ($confidence eq 'unknown') {
+                                 $confidence = -1;
+                             }
+                             $loadIsPossiblePlaceFor->Put($psort, $featureID, $confidence);
+                             # If this cell location does not yet exist, create its record.
+                             if (! exists $cellLocation{$psort}) {
+                                 $cellLocation{$psort} = 1;
+                                 $loadCellLocation->Put($psort);
+                             }
+                             # If this is a significant location, add it as a keyword.
+                             if ($confidence > 2.5) {
+                                 push @keywords, $psort;
+                             }
+                         }
+                     }
+                     # Phobius data is next. This consists of the signal peptide location and
+                     # the transmembrane locations.
+                     my $signalList = "";
+                     my $transList = "";
+                     if (exists $attributeHash{Phobius}) {
+                         # This will be a hash of two keys (transmembrane and signal) to
+                         # location strings. If there's no value, we stuff in an empty string.
+                         $signalList = ($attributeHash{Phobius}->{signal} || "");
+                         $transList = ($attributeHash{Phobius}->{transmembrane} || "");
+                     }
+                     # Here are some more numbers: isoelectric point, molecular weight, and
+                     # the similar-to-human flag.
+                     my $isoelectric = 0;
+                     if (exists $attributeHash{isoelectric_point}) {
+                         $isoelectric = $attributeHash{isoelectric_point}->{""};
+                     }
+                     my $similarToHuman = 0;
+                     if (exists $attributeHash{similar_to_human} && $attributeHash{similar_to_human}->{""} eq 'yes') {
+                         $similarToHuman = 1;
+                     }
+                     my $molecularWeight = 0;
+                     if (exists $attributeHash{molecular_weight}) {
+                         $molecularWeight = $attributeHash{molecular_weight}->{""};
+                     }
+                     # Create the keyword string.
+                     my $keywordString = join(" ", @keywords);
+                     Trace("Real keyword string for $featureID: $keywordString.") if T(4);
+                     # Get rid of annoying punctuation.
+                     $keywordString =~ s/[();@#\/]/ /g;
+                     # Get the list of keywords in the keyword string.
+                     my @realKeywords = grep { $biowords->IsWord($_) } $biowords->Split($keywordString);
+                     # We need to do two things here: create the keyword string for the feature table
+                     # and write records to the keyword and stem files. The stuff we write to
+                     # the files will be taken from the following two hashes. The stuff used
+                     # to create the keyword string will be taken from the list.
+                     my (%keys, %stems, @realStems);
+                     for my $keyword (@realKeywords) {
+                         # Compute the stem and phonex for this keyword.
+                         my ($stem, $phonex) = $biowords->StemLookup($keyword);
+                         # Only proceed if a stem comes back. If no stem came back, it's a
+                         # stop word and we throw it away.
+                         if ($stem) {
+                             $keys{$keyword} = $stem;
+                             $stems{$stem} = $phonex;
+                             push @realStems, $stem;
+                         }
+                     }
+                     # Now create the keyword string.
+                     my $cleanWords = join(" ", @realStems);
+                     Trace("Keyword string for $featureID: $cleanWords") if T(4);
+                     # Write the stem and keyword records.
+                     for my $stem (keys %stems) {
+                         Tracer::PutLine($stemh, [$stem, $stems{$stem}]);
+                     }
+                     for my $key (keys %keys) {
+                         # The stem goes first in this file, because we want to sort
+                         # by stem and then keyword.
+                         Tracer::PutLine($keyh, [$keys{$key}, $key]);
+                     }
+                     # Now we need to process the feature's locations. First, we split them up.
+                     my @locationList = split /\s*,\s*/, $locations;
+                     # Next, we convert them to Sprout location objects.
+                     my @locObjectList = map { BasicLocation->new("$genomeID:$_") } @locationList;
+                     # Assemble them into a sprout location string for later.
+                     my $locationString = join(", ", map { $_->String } @locObjectList);
+                     # We'll store the sequence length in here.
+                     my $sequenceLength = 0;
                  # This part is the roughest. We need to relate the features to contig
                  # locations, and the locations must be split so that none of them exceed
                  # the maximum segment size. This simplifies the genes_in_region processing
-                 # for Sprout.
+                     # for Sprout. To start, we create the location position indicator.
-                 my @locationList = split /\s*,\s*/, $locations;
-                 # Create the location position indicator.
                  my $i = 1;
                  # Loop through the locations.
-                 for my $location (@locationList) {
+                     for my $locObject (@locObjectList) {
-                     # Parse the location.
+                         # Record the length.
-                     my $locObject = BasicLocation->new("$genomeID:$location");
+                         $sequenceLength += $locObject->Length;
-                     # Split it into a list of chunks.
+                         # Split this location into a list of chunks.
                      my @locOList = ();
                      while (my $peeling = $locObject->Peel($chunkSize)) {
                          $loadIsLocatedIn->Add("peeling");
-Line 529
+Line 741
                          $i++;
                      }
                  }
-             }
+                     # Now we get some ancillary flags.
-         }
+                     my $locked = $fig->is_locked_fid($featureID);
-     }
+                     my $in_genbank = $fig->peg_in_gendb($featureID);
-     # Finish the loads.
+                     # Create the feature record.
-     my $retVal = $self->_FinishAll();
+                     $loadFeature->Put($featureID, 1, $user, $quality, $type, $in_genbank, $isoelectric, $locked, $molecularWeight,
-     return $retVal;
+                                       $sequenceLength, $signalList, $similarToHuman, $assignment, $cleanWords, $locationString,
- }
+                                       $transList);
+                 }
- =head3 LoadBBHData
+             }
+             Trace("Genome $genomeID processed.") if T(3);
- C<< my $stats = $spl->LoadBBHData(); >>
+         }
+     }
- Load the bidirectional best hit data from FIG into Sprout.
+     Trace("Sorting keywords.") if T(2);
+     # Now we need to load the keyword table from the key and stem files.
- Sprout does not store information on similarities. Instead, it has only the
+     close $keyh;
- bi-directional best hits. Even so, the BBH table is one of the largest in
+     close $stemh;
- the database.
+     Trace("Loading keywords.") if T(2);
+     $keyh = Open(undef, "<$keyFileName");
- The following relations are loaded by this method.
+     $stemh = Open(undef, "<$stemFileName");
+     # We'll count the keywords in here, for tracing purposes.
-     IsBidirectionalBestHitOf
+     my $count = 0;
+     # These variables track the current stem's data. When an incoming
- =over 4
+     # keyword's stem changes, these will be recomputed.
+     my ($currentStem, $currentPhonex, $currentCount);
- =item RETURNS
+     # Prime the loop by reading the first stem in the stem file.
+     my ($nextStem, $nextPhonex) = Tracer::GetLine($stemh);
- Returns a statistics object for the loads.
+     # Loop through the keyword file.
+     while (! eof $keyh) {
- =back
+         # Read this keyword.
+         my ($thisStem, $thisKey) = Tracer::GetLine($keyh);
- =cut
+         # Check to see if it's the new stem yet.
- #: Return Type $%;
+         if ($thisStem ne $currentStem) {
- sub LoadBBHData {
+             # Yes. It's a terrible error if it's not also the next stem.
-     # Get this object instance.
+             if ($thisStem ne $nextStem) {
-     my ($self) = @_;
+                 Confess("Error in stem file. Expected \"$nextStem\", but found \"$thisStem\".");
-     # Get the FIG object.
-     my $fig = $self->{fig};
-     # Get the table of genome IDs.
-     my $genomeHash = $self->{genomes};
-     # Create load objects for each of the tables we're loading.
-     my $loadIsBidirectionalBestHitOf = $self->_TableLoader('IsBidirectionalBestHitOf');
-     if ($self->{options}->{loadOnly}) {
-         Trace("Loading from existing files.") if T(2);
      } else {
-         Trace("Generating BBH data.") if T(2);
+                 # Here we're okay.
-         # Now we loop through the genomes, generating the data for each one.
+                 ($currentStem, $currentPhonex) = ($nextStem, $nextPhonex);
-         for my $genomeID (sort keys %{$genomeHash}) {
+                 # Count the number of features for this stem.
-             $loadIsBidirectionalBestHitOf->Add("genomeIn");
+                 $currentCount = 0;
-             Trace("Processing features for genome $genomeID.") if T(3);
+                 while ($nextStem eq $thisStem) {
-             # Get the feature list for this genome.
+                     ($nextStem, $nextPhonex) = Tracer::GetLine($stemh);
-             my $features = $fig->all_features_detailed($genomeID);
+                     $currentCount++;
-             # Loop through the features.
-             for my $featureData (@{$features}) {
-                 # Split the tuple.
-                 my ($featureID, $locations, $aliases, $type) = @{$featureData};
-                 # Get the bi-directional best hits.
-                 my @bbhList = $fig->bbhs($featureID);
-                 for my $bbhEntry (@bbhList) {
-                     # Get the target feature ID and the score.
-                     my ($targetID, $score) = @{$bbhEntry};
-                     # Check the target feature's genome.
-                     my $targetGenomeID = $fig->genome_of($targetID);
-                     # Only proceed if it's one of our genomes.
-                     if ($genomeHash->{$targetGenomeID}) {
-                         $loadIsBidirectionalBestHitOf->Put($featureID, $targetID, $targetGenomeID,
-                                                            $score);
                      }
                  }
              }
+         # Now $currentStem is the same as $thisStem, and the other $current-vars
+         # contain the stem's data (phonex and count).
+         $loadKeyword->Put($thisKey, $currentCount, $currentPhonex, $currentStem);
+         if (++$count % 1000 == 0 && T(3)) {
+             Trace("$count keywords loaded.");
          }
      }
+     Trace("$count keywords loaded into keyword table.") if T(2);
      # Finish the loads.
      my $retVal = $self->_FinishAll();
      return $retVal;
-Line 607
+Line 802
  =head3 LoadSubsystemData
- C<< my $stats = $spl->LoadSubsystemData(); >>
+     my $stats = $spl->LoadSubsystemData();
  Load the subsystem data from FIG into Sprout.
-Line 620
+Line 815
  The following relations are loaded by this method.
      Subsystem
+     SubsystemClass
      Role
      RoleEC
+     IsIdentifiedByEC
      SSCell
      ContainsFeature
      IsGenomeOf
-Line 635
+Line 832
      ConsistsOfGenomes
      GenomeSubset
      HasGenomeSubset
-     Catalyzes
      Diagram
      RoleOccursIn
+     SubsystemHopeNotes
  =over 4
-Line 663
+Line 860
      # Get the map list.
      my @maps = $fig->all_maps;
      # Create load objects for each of the tables we're loading.
-     my $loadDiagram = $self->_TableLoader('Diagram', $self->PrimaryOnly);
+     my $loadDiagram = $self->_TableLoader('Diagram');
-     my $loadRoleOccursIn = $self->_TableLoader('RoleOccursIn', $self->PrimaryOnly);
+     my $loadRoleOccursIn = $self->_TableLoader('RoleOccursIn');
      my $loadSubsystem = $self->_TableLoader('Subsystem');
-     my $loadRole = $self->_TableLoader('Role', $self->PrimaryOnly);
+     my $loadRole = $self->_TableLoader('Role');
-     my $loadRoleEC = $self->_TableLoader('RoleEC', $self->PrimaryOnly);
+     my $loadRoleEC = $self->_TableLoader('RoleEC');
-     my $loadCatalyzes = $self->_TableLoader('Catalyzes', $self->PrimaryOnly);
+     my $loadIsIdentifiedByEC = $self->_TableLoader('IsIdentifiedByEC');
-     my $loadSSCell = $self->_TableLoader('SSCell', $self->PrimaryOnly);
+     my $loadCatalyzes = $self->_TableLoader('Catalyzes');
-     my $loadContainsFeature = $self->_TableLoader('ContainsFeature', $self->PrimaryOnly);
+     my $loadSSCell = $self->_TableLoader('SSCell');
-     my $loadIsGenomeOf = $self->_TableLoader('IsGenomeOf', $self->PrimaryOnly);
+     my $loadContainsFeature = $self->_TableLoader('ContainsFeature');
-     my $loadIsRoleOf = $self->_TableLoader('IsRoleOf', $self->PrimaryOnly);
+     my $loadIsGenomeOf = $self->_TableLoader('IsGenomeOf');
-     my $loadOccursInSubsystem = $self->_TableLoader('OccursInSubsystem', $self->PrimaryOnly);
+     my $loadIsRoleOf = $self->_TableLoader('IsRoleOf');
-     my $loadParticipatesIn = $self->_TableLoader('ParticipatesIn', $self->PrimaryOnly);
+     my $loadOccursInSubsystem = $self->_TableLoader('OccursInSubsystem');
-     my $loadHasSSCell = $self->_TableLoader('HasSSCell', $self->PrimaryOnly);
+     my $loadParticipatesIn = $self->_TableLoader('ParticipatesIn');
-     my $loadRoleSubset = $self->_TableLoader('RoleSubset', $self->PrimaryOnly);
+     my $loadHasSSCell = $self->_TableLoader('HasSSCell');
-     my $loadGenomeSubset = $self->_TableLoader('GenomeSubset', $self->PrimaryOnly);
+     my $loadRoleSubset = $self->_TableLoader('RoleSubset');
-     my $loadConsistsOfRoles = $self->_TableLoader('ConsistsOfRoles', $self->PrimaryOnly);
+     my $loadGenomeSubset = $self->_TableLoader('GenomeSubset');
-     my $loadConsistsOfGenomes = $self->_TableLoader('ConsistsOfGenomes', $self->PrimaryOnly);
+     my $loadConsistsOfRoles = $self->_TableLoader('ConsistsOfRoles');
-     my $loadHasRoleSubset = $self->_TableLoader('HasRoleSubset', $self->PrimaryOnly);
+     my $loadConsistsOfGenomes = $self->_TableLoader('ConsistsOfGenomes');
-     my $loadHasGenomeSubset = $self->_TableLoader('HasGenomeSubset', $self->PrimaryOnly);
+     my $loadHasRoleSubset = $self->_TableLoader('HasRoleSubset');
+     my $loadHasGenomeSubset = $self->_TableLoader('HasGenomeSubset');
+     my $loadSubsystemClass = $self->_TableLoader('SubsystemClass');
+     my $loadSubsystemHopeNotes = $self->_TableLoader('SubsystemHopeNotes');
      if ($self->{options}->{loadOnly}) {
          Trace("Loading from existing files.") if T(2);
      } else {
          Trace("Generating subsystem data.") if T(2);
-         # This hash will contain the role for each EC. When we're done, this
+         # This hash will contain the roles for each EC. When we're done, this
          # information will be used to generate the Catalyzes table.
          my %ecToRoles = ();
          # Loop through the subsystems. Our first task will be to create the
-Line 697
+Line 897
          my ($genomeID, $roleID);
          my %roleData = ();
          for my $subsysID (@subsysIDs) {
-             Trace("Creating subsystem $subsysID.") if T(3);
-             $loadSubsystem->Add("subsystemIn");
              # Get the subsystem object.
              my $sub = $fig->get_subsystem($subsysID);
+             # Only proceed if the subsystem has a spreadsheet.
+             if (defined($sub) && ! $sub->{empty_ss}) {
+                 Trace("Creating subsystem $subsysID.") if T(3);
+                 $loadSubsystem->Add("subsystemIn");
              # Create the subsystem record.
              my $curator = $sub->get_curator();
              my $notes = $sub->get_notes();
-             $loadSubsystem->Put($subsysID, $curator, $notes);
+                 my $version = $sub->get_version();
+                 my $description = $sub->get_description();
+                 $loadSubsystem->Put($subsysID, $curator, $version, $description, $notes);
+                 # Add the hope notes.
+                 my $hopeNotes = $sub->get_hope_curation_notes();
+                 if ($hopeNotes) {
+                     $loadSubsystemHopeNotes->Put($sub, $hopeNotes);
+                 }
+                 # Now for the classification string. This comes back as a list
+                 # reference and we convert it to a space-delimited string.
+                 my $classList = $fig->subsystem_classification($subsysID);
+                 my $classString = join($FIG_Config::splitter, grep { $_ } @$classList);
+                 $loadSubsystemClass->Put($subsysID, $classString);
              # Connect it to its roles. Each role is a column in the subsystem spreadsheet.
              for (my $col = 0; defined($roleID = $sub->get_role($col)); $col++) {
+                     # Get the role's abbreviation.
+                     my $abbr = $sub->get_role_abbr($col);
+                     # Get its essentiality.
+                     my $aux = $fig->is_aux_role_in_subsystem($subsysID, $roleID);
+                     # Get its reaction note.
+                     my $hope_note = $sub->get_hope_reaction_notes($roleID) || "";
                  # Connect to this role.
                  $loadOccursInSubsystem->Add("roleIn");
-                 $loadOccursInSubsystem->Put($roleID, $subsysID, $col);
+                     $loadOccursInSubsystem->Put($roleID, $subsysID, $abbr, $aux, $col, $hope_note);
                  # If it's a new role, add it to the role table.
                  if (! exists $roleData{$roleID}) {
                      # Get the role's abbreviation.
-                     my $abbr = $sub->get_role_abbr($col);
                      # Add the role.
-                     $loadRole->Put($roleID, $abbr);
+                         $loadRole->Put($roleID);
                      $roleData{$roleID} = 1;
                      # Check for an EC number.
-                     if ($roleID =~ /\(EC ([^.]+\.[^.]+\.[^.]+\.[^)]+)\)\s*$/) {
+                         if ($roleID =~ /\(EC (\d+\.\d+\.\d+\.\d+)\s*\)\s*$/) {
                          my $ec = $1;
-                         $loadRoleEC->Put($roleID, $ec);
+                             $loadIsIdentifiedByEC->Put($roleID, $ec);
-                         $ecToRoles{$ec} = $roleID;
+                             # Check to see if this is our first encounter with this EC.
+                             if (exists $ecToRoles{$ec}) {
+                                 # No, so just add this role to the EC list.
+                                 push @{$ecToRoles{$ec}}, $roleID;
+                             } else {
+                                 # Output this EC.
+                                 $loadRoleEC->Put($ec);
+                                 # Create its role list.
+                                 $ecToRoles{$ec} = [$roleID];
+                             }
                      }
                  }
              }
-Line 748
+Line 976
                      # part of the spreadsheet cell ID.
                      for (my $col = 0; defined($roleID = $sub->get_role($col)); $col++) {
                          # Get the features in the spreadsheet cell for this genome and role.
-                         my @pegs = $sub->get_pegs_from_cell($row, $col);
+                             my @pegs = grep { !$fig->is_deleted_fid($_) } $sub->get_pegs_from_cell($row, $col);
                          # Only proceed if features exist.
                          if (@pegs > 0) {
                              # Create the spreadsheet cell.
-Line 769
+Line 997
                      if ($pegCount > 0) {
                          Trace("$pegCount PEGs in $cellCount cells for $genomeID.") if T(3);
                          $loadParticipatesIn->Put($genomeID, $subsysID, $variantCode);
-                         # Partition the PEGs found into clusters.
-                         my @clusters = $fig->compute_clusters(\@pegsFound, $sub);
                          # Create a hash mapping PEG IDs to cluster numbers.
                          # We default to -1 for all of them.
                          my %clusterOf = map { $_ => -1 } @pegsFound;
+                             # Partition the PEGs found into clusters.
+                             my @clusters = $fig->compute_clusters([keys %clusterOf], $sub);
                          for (my $i = 0; $i <= $#clusters; $i++) {
                              my $subList = $clusters[$i];
                              for my $peg (@{$subList}) {
-Line 801
+Line 1029
                  # Connect the subset to the subsystem.
                  $loadHasRoleSubset->Put($subsysID, $actualID);
                  # Connect the subset to its roles.
-                 my @roles = $sub->get_subset($subsetID);
+                     my @roles = $sub->get_subsetC_roles($subsetID);
                  for my $roleID (@roles) {
                      $loadConsistsOfRoles->Put($actualID, $roleID);
                  }
-Line 821
+Line 1049
                  }
              }
          }
+         }
          # Now we loop through the diagrams. We need to create the diagram records
          # and link each diagram to its roles. Note that only roles which occur
          # in subsystems (and therefore appear in the %ecToRoles hash) are
-Line 833
+Line 1062
              # Now we need to link all the map's roles to it.
              # A hash is used to prevent duplicates.
              my %roleHash = ();
-             for my $role ($fig->map_to_ecs($map)) {
+             for my $ec ($fig->map_to_ecs($map)) {
-                 if (exists $ecToRoles{$role} && ! $roleHash{$role}) {
+                 if (exists $ecToRoles{$ec}) {
-                     $loadRoleOccursIn->Put($ecToRoles{$role}, $map);
+                     for my $role (@{$ecToRoles{$ec}}) {
+                         if (! $roleHash{$role}) {
+                             $loadRoleOccursIn->Put($role, $map);
                      $roleHash{$role} = 1;
                  }
              }
          }
-         # Before we leave, we must create the Catalyzes table. We start with the reactions,
-         # then use the "ecToRoles" table to convert EC numbers to role IDs.
-         my @reactions = $fig->all_reactions();
-         for my $reactionID (@reactions) {
-             # Get this reaction's list of roles. The results will be EC numbers.
-             my @roles = $fig->catalyzed_by($reactionID);
-             # Loop through the roles, creating catalyzation records.
-             for my $thisRole (@roles) {
-                 if (exists $ecToRoles{$thisRole}) {
-                     $loadCatalyzes->Put($ecToRoles{$thisRole}, $reactionID);
-                 }
              }
          }
      }
-Line 861
+Line 1081
  =head3 LoadPropertyData
- C<< my $stats = $spl->LoadPropertyData(); >>
+     my $stats = $spl->LoadPropertyData();
  Load the attribute data from FIG into Sprout.
-Line 897
+Line 1117
      my $genomeHash = $self->{genomes};
      # Create load objects for each of the tables we're loading.
      my $loadProperty = $self->_TableLoader('Property');
-     my $loadHasProperty = $self->_TableLoader('HasProperty', $self->PrimaryOnly);
+     my $loadHasProperty = $self->_TableLoader('HasProperty');
      if ($self->{options}->{loadOnly}) {
          Trace("Loading from existing files.") if T(2);
      } else {
-Line 905
+Line 1125
          # Create a hash for storing property IDs.
          my %propertyKeys = ();
          my $nextID = 1;
+         # Get the attributes we intend to store in the property table.
+         my $propKeys = $self->{propKeys};
          # Loop through the genomes.
-         for my $genomeID (keys %{$genomeHash}) {
+         for my $genomeID (sort keys %{$genomeHash}) {
              $loadProperty->Add("genomeIn");
              Trace("Generating properties for $genomeID.") if T(3);
-             # Get the genome's features. The feature ID is the first field in the
+             # Initialize a counter.
-             # tuples returned by "all_features_detailed". We use "all_features_detailed"
-             # rather than "all_features" because we want all features regardless of type.
-             my @features = map { $_->[0] } @{$fig->all_features_detailed($genomeID)};
-             my $featureCount = 0;
              my $propertyCount = 0;
-             # Loop through the features, creating HasProperty records.
+             # Get the properties for this genome's features.
-             for my $fid (@features) {
+             my @attributes = $fig->get_attributes("fig|$genomeID%", $propKeys);
-                 # Get all attributes for this feature. We do this one feature at a time
+             Trace("Property list built for $genomeID.") if T(3);
-                 # to insure we do not get any genome attributes.
+             # Loop through the results, creating HasProperty records.
-                 my @attributeList = $fig->get_attributes($fid, '', '', '');
+             for my $attributeData (@attributes) {
-                 if (scalar @attributeList) {
+                 # Pull apart the attribute tuple.
-                     $featureCount++;
+                 my ($fid, $key, $value, $url) = @{$attributeData};
-                 }
-                 # Loop through the attributes.
-                 for my $tuple (@attributeList) {
-                     $propertyCount++;
-                     # Get this attribute value's data. Note that we throw away the FID,
-                     # since it will always be the same as the value if "$fid".
-                     my (undef, $key, $value, $url) = @{$tuple};
                      # Concatenate the key and value and check the "propertyKeys" hash to
                      # see if we already have an ID for it. We use a tab for the separator
                      # character.
-Line 946
+Line 1157
                      # Create the HasProperty entry for this feature/property association.
                      $loadHasProperty->Put($fid, $propertyID, $url);
                  }
-             }
              # Update the statistics.
-             Trace("$propertyCount attributes processed for $featureCount features.") if T(3);
+             Trace("$propertyCount attributes processed.") if T(3);
-             $loadHasProperty->Add("featuresIn", $featureCount);
              $loadHasProperty->Add("propertiesIn", $propertyCount);
          }
      }
-Line 960
+Line 1169
  =head3 LoadAnnotationData
- C<< my $stats = $spl->LoadAnnotationData(); >>
+     my $stats = $spl->LoadAnnotationData();
  Load the annotation data from FIG into Sprout.
-Line 994
+Line 1203
      my $genomeHash = $self->{genomes};
      # Create load objects for each of the tables we're loading.
      my $loadAnnotation = $self->_TableLoader('Annotation');
-     my $loadIsTargetOfAnnotation = $self->_TableLoader('IsTargetOfAnnotation', $self->PrimaryOnly);
+     my $loadIsTargetOfAnnotation = $self->_TableLoader('IsTargetOfAnnotation');
-     my $loadSproutUser = $self->_TableLoader('SproutUser', $self->PrimaryOnly);
+     my $loadSproutUser = $self->_TableLoader('SproutUser');
-     my $loadUserAccess = $self->_TableLoader('UserAccess', $self->PrimaryOnly);
+     my $loadUserAccess = $self->_TableLoader('UserAccess');
-     my $loadMadeAnnotation = $self->_TableLoader('MadeAnnotation', $self->PrimaryOnly);
+     my $loadMadeAnnotation = $self->_TableLoader('MadeAnnotation');
      if ($self->{options}->{loadOnly}) {
          Trace("Loading from existing files.") if T(2);
      } else {
-Line 1015
+Line 1224
          # Loop through the genomes.
          for my $genomeID (sort keys %{$genomeHash}) {
              Trace("Processing $genomeID.") if T(3);
-             # Get the genome's PEGs.
-             my @pegs = $fig->pegs_of($genomeID);
-             for my $peg (@pegs) {
-                 Trace("Processing $peg.") if T(4);
                  # Create a hash of timestamps. We use this to prevent duplicate time stamps
                  # from showing up for a single PEG's annotations.
                  my %seenTimestamps = ();
-                 # Loop through the annotations.
+             # Get the genome's annotations.
-                 for my $tuple ($fig->feature_annotations($peg, "raw")) {
+             my @annotations = $fig->read_all_annotations($genomeID);
-                     my ($fid, $timestamp, $user, $text) = @{$tuple};
+             Trace("Processing annotations.") if T(2);
+             for my $tuple (@annotations) {
+                 # Get the annotation tuple.
+                 my ($peg, $timestamp, $user, $text) = @{$tuple};
                      # Here we fix up the annotation text. "\r" is removed,
-                     # and "\t" and "\n" are escaped. Note we use the "s"
+                 # and "\t" and "\n" are escaped. Note we use the "gs"
                      # modifier so that new-lines inside the text do not
                      # stop the substitution search.
                      $text =~ s/\r//gs;
-Line 1039
+Line 1247
                          # Here it's a number. We need to insure the one we use to form
                          # the key is unique.
                          my $keyStamp = $timestamp;
-                         while ($seenTimestamps{$keyStamp}) {
+                     while ($seenTimestamps{"$peg:$keyStamp"}) {
                              $keyStamp++;
                          }
-                         $seenTimestamps{$keyStamp} = 1;
                          my $annotationID = "$peg:$keyStamp";
+                     $seenTimestamps{$annotationID} = 1;
                          # Insure the user exists.
                          if (! $users{$user}) {
                              $loadSproutUser->Put($user, "SEED user");
-Line 1061
+Line 1269
                  }
              }
          }
-     }
      # Finish the load.
      my $retVal = $self->_FinishAll();
      return $retVal;
-Line 1069
+Line 1276
  =head3 LoadSourceData
- C<< my $stats = $spl->LoadSourceData(); >>
+     my $stats = $spl->LoadSourceData();
  Load the source data from FIG into Sprout.
-Line 1103
+Line 1310
      # Get the genome hash.
      my $genomeHash = $self->{genomes};
      # Create load objects for each of the tables we're loading.
-     my $loadComesFrom = $self->_TableLoader('ComesFrom', $self->PrimaryOnly);
+     my $loadComesFrom = $self->_TableLoader('ComesFrom');
      my $loadSource = $self->_TableLoader('Source');
      my $loadSourceURL = $self->_TableLoader('SourceURL');
      if ($self->{options}->{loadOnly}) {
-Line 1147
+Line 1354
  =head3 LoadExternalData
- C<< my $stats = $spl->LoadExternalData(); >>
+     my $stats = $spl->LoadExternalData();
  Load the external data from FIG into Sprout.
-Line 1190
+Line 1397
      } else {
          Trace("Generating external data.") if T(2);
          # We loop through the files one at a time. First, the organism file.
-         Open(\*ORGS, "<$FIG_Config::global/ext_org.table");
+         Open(\*ORGS, "sort +0 -1 -u -t\"\t\" $FIG_Config::global/ext_org.table |");
          my $orgLine;
          while (defined($orgLine = <ORGS>)) {
              # Clean the input line.
-Line 1202
+Line 1409
          close ORGS;
          # Now the function file.
          my $funcLine;
-         Open(\*FUNCS, "<$FIG_Config::global/ext_func.table");
+         Open(\*FUNCS, "sort +0 -1 -u -t\"\t\" $FIG_Config::global/ext_func.table |");
          while (defined($funcLine = <FUNCS>)) {
              # Clean the line ending.
              chomp $funcLine;
-Line 1227
+Line 1434
  =head3 LoadReactionData
- C<< my $stats = $spl->LoadReactionData(); >>
+     my $stats = $spl->LoadReactionData();
  Load the reaction data from FIG into Sprout.
-Line 1240
+Line 1447
      Compound
      CompoundName
      CompoundCAS
+     IsIdentifiedByCAS
+     HasCompoundName
      IsAComponentOf
+     Scenario
+     Catalyzes
+     HasScenario
+     IsInputFor
+     IsOutputOf
+     ExcludesReaction
+     IncludesReaction
+     IsOnDiagram
+     IncludesReaction
  This method proceeds reaction by reaction rather than genome by genome.
-Line 1261
+Line 1479
      my $fig = $self->{fig};
      # Create load objects for each of the tables we're loading.
      my $loadReaction = $self->_TableLoader('Reaction');
-     my $loadReactionURL = $self->_TableLoader('ReactionURL', $self->PrimaryOnly);
+     my $loadReactionURL = $self->_TableLoader('ReactionURL');
-     my $loadCompound = $self->_TableLoader('Compound', $self->PrimaryOnly);
+     my $loadCompound = $self->_TableLoader('Compound');
-     my $loadCompoundName = $self->_TableLoader('CompoundName', $self->PrimaryOnly);
+     my $loadCompoundName = $self->_TableLoader('CompoundName');
-     my $loadCompoundCAS = $self->_TableLoader('CompoundCAS', $self->PrimaryOnly);
+     my $loadCompoundCAS = $self->_TableLoader('CompoundCAS');
-     my $loadIsAComponentOf = $self->_TableLoader('IsAComponentOf', $self->PrimaryOnly);
+     my $loadIsAComponentOf = $self->_TableLoader('IsAComponentOf');
+     my $loadIsIdentifiedByCAS = $self->_TableLoader('IsIdentifiedByCAS');
+     my $loadHasCompoundName = $self->_TableLoader('HasCompoundName');
+     my $loadScenario = $self->_TableLoader('Scenario');
+     my $loadHasScenario = $self->_TableLoader('HasScenario');
+     my $loadIsInputFor = $self->_TableLoader('IsInputFor');
+     my $loadIsOutputOf = $self->_TableLoader('IsOutputOf');
+     my $loadIsOnDiagram = $self->_TableLoader('IsOnDiagram');
+     my $loadIncludesReaction = $self->_TableLoader('IncludesReaction');
+     my $loadExcludesReaction = $self->_TableLoader('ExcludesReaction');
+     my $loadCatalyzes = $self->_TableLoader('Catalyzes');
      if ($self->{options}->{loadOnly}) {
          Trace("Loading from existing files.") if T(2);
      } else {
-         Trace("Generating annotation data.") if T(2);
+         Trace("Generating reaction data.") if T(2);
+         # We need some hashes to prevent duplicates.
+         my %compoundNames = ();
+         my %compoundCASes = ();
          # First we create the compounds.
-         my @compounds = $fig->all_compounds();
+         my %compounds = map { $_ => 1 } $fig->all_compounds();
-         for my $cid (@compounds) {
+         for my $cid (keys %compounds) {
              # Check for names.
              my @names = $fig->names_of_compound($cid);
              # Each name will be given a priority number, starting with 1.
              my $prio = 1;
              for my $name (@names) {
-                 $loadCompoundName->Put($cid, $name, $prio++);
+                 if (! exists $compoundNames{$name}) {
+                     $loadCompoundName->Put($name);
+                     $compoundNames{$name} = 1;
+                 }
+                 $loadHasCompoundName->Put($cid, $name, $prio++);
              }
              # Create the main compound record. Note that the first name
              # becomes the label.
-Line 1287
+Line 1522
              # Check for a CAS ID.
              my $cas = $fig->cas($cid);
              if ($cas) {
-                 $loadCompoundCAS->Put($cid, $cas);
+                 $loadIsIdentifiedByCAS->Put($cid, $cas);
+                 if (! exists $compoundCASes{$cas}) {
+                     $loadCompoundCAS->Put($cas);
+                     $compoundCASes{$cas} = 1;
+                 }
              }
          }
          # All the compounds are set up, so we need to loop through the reactions next. First,
          # we initialize the discriminator index. This is a single integer used to insure
          # duplicate elements in a reaction are not accidentally collapsed.
          my $discrim = 0;
-         my @reactions = $fig->all_reactions();
+         my %reactions = map { $_ => 1 } $fig->all_reactions();
-         for my $reactionID (@reactions) {
+         for my $reactionID (keys %reactions) {
              # Create the reaction record.
              $loadReaction->Put($reactionID, $fig->reversible($reactionID));
              # Compute the reaction's URL.
-Line 1318
+Line 1557
                  }
              }
          }
+         # Now we run through the subsystems and roles, generating the scenarios
+         # and connecting the reactions. We'll need some hashes to prevent
+         # duplicates and a counter for compound group keys.
+         my %roles = ();
+         my %scenarios = ();
+         my @subsystems = $fig->all_subsystems();
+         for my $subName (@subsystems) {
+             my $sub = $fig->get_subsystem($subName);
+             Trace("Processing $subName reactions.") if T(3);
+             # Get the subsystem's reactions.
+             my %reactions = $sub->get_hope_reactions();
+             # Loop through the roles, connecting them to the reactions.
+             for my $role (keys %reactions) {
+                 # Only process this role if it is new.
+                 if (! $roles{$role}) {
+                     $roles{$role} = 1;
+                     my @reactions = @{$reactions{$role}};
+                     for my $reaction (@reactions) {
+                         $loadCatalyzes->Put($role, $reaction);
+                     }
+                 }
+             }
+             Trace("Processing $subName scenarios.") if T(3);
+             # Get the subsystem's scenarios.
+             my @scenarioNames = $sub->get_hope_scenario_names();
+             # Loop through the scenarios, creating scenario data.
+             for my $scenarioName (@scenarioNames) {
+                 # Link this scenario to this subsystem.
+                 $loadHasScenario->Put($subName, $scenarioName);
+                 # If this scenario is new, we need to create it.
+                 if (! $scenarios{$scenarioName}) {
+                     Trace("Creating scenario $scenarioName.") if T(3);
+                     $scenarios{$scenarioName} = 1;
+                     # Create the scenario itself.
+                     $loadScenario->Put($scenarioName);
+                     # Attach the input compounds.
+                     for my $input ($sub->get_hope_input_compounds($scenarioName)) {
+                         $loadIsInputFor->Put($input, $scenarioName);
+                     }
+                     # Now we need to set up the output compounds. They come in two
+                     # groups, which we mark 0 and 1.
+                     my $outputGroup = 0;
+                     # Set up the output compounds.
+                     for my $outputGroup ($sub->get_hope_output_compounds($scenarioName)) {
+                         # Attach the compounds.
+                         for my $compound (@$outputGroup) {
+                             $loadIsOutputOf->Put($scenarioName, $compound, $outputGroup);
+                         }
+                     }
+                     # Create the reaction lists.
+                     my @addReactions = $sub->get_hope_additional_reactions($scenarioName);
+                     for my $reaction (@addReactions) {
+                         $loadIncludesReaction->Put($scenarioName, $reaction);
+                     }
+                     my @notReactions = $sub->get_hope_ignore_reactions($scenarioName);
+                     for my $reaction (@notReactions) {
+                         $loadExcludesReaction->Put($scenarioName, $reaction);
+                     }
+                     # Link the maps.
+                     my @maps = $sub->get_hope_map_ids($scenarioName);
+                     for my $map (@maps) {
+                         $loadIsOnDiagram->Put($scenarioName, "map$map");
+                     }
+                 }
+             }
+         }
      }
      # Finish the load.
      my $retVal = $self->_FinishAll();
      return $retVal;
  }
- =head3 LoadGroupData
+ =head3 LoadSynonymData
- C<< my $stats = $spl->LoadGroupData(); >>
+     my $stats = $spl->LoadSynonymData();
- Load the genome Groups into Sprout.
+ Load the synonym groups into Sprout.
  The following relations are loaded by this method.
-     GenomeGroups
+     SynonymGroup
+     IsSynonymGroupFor
- There is no direct support for genome groups in FIG, so we access the SEED
+ The source information for these relations is taken from the C<maps_to_id> method
- files directly.
+ of the B<FIG> object. Unfortunately, to make this work, we need to use direct
+ SQL against the FIG database.
  =over 4
-Line 1347
+Line 1654
  =cut
  #: Return Type $%;
- sub LoadGroupData {
+ sub LoadSynonymData {
      # Get this object instance.
      my ($self) = @_;
      # Get the FIG object.
-Line 1355
+Line 1662
      # Get the genome hash.
      my $genomeHash = $self->{genomes};
      # Create a load object for the table we're loading.
-     my $loadGenomeGroups = $self->_TableLoader('GenomeGroups');
+     my $loadSynonymGroup = $self->_TableLoader('SynonymGroup');
+     my $loadIsSynonymGroupFor = $self->_TableLoader('IsSynonymGroupFor');
+     if ($self->{options}->{loadOnly}) {
+         Trace("Loading from existing files.") if T(2);
+     } else {
+         Trace("Generating synonym group data.") if T(2);
+         # Get the database handle.
+         my $dbh = $fig->db_handle();
+         # Ask for the synonyms. Note that "maps_to" is a group name, and "syn_id" is a PEG ID or alias.
+         my $sth = $dbh->prepare_command("SELECT maps_to, syn_id FROM peg_synonyms ORDER BY maps_to");
+         my $result = $sth->execute();
+         if (! defined($result)) {
+             Confess("Database error in Synonym load: " . $sth->errstr());
+         } else {
+             Trace("Processing synonym results.") if T(2);
+             # Remember the current synonym.
+             my $current_syn = "";
+             # Count the features.
+             my $featureCount = 0;
+             my $entryCount = 0;
+             # Loop through the synonym/peg pairs.
+             while (my @row = $sth->fetchrow()) {
+                 # Get the synonym group ID and feature ID.
+                 my ($syn_id, $peg) = @row;
+                 # Count this row.
+                 $entryCount++;
+                 if ($entryCount % 1000 == 0) {
+                     Trace("$entryCount rows processed.") if T(3);
+                 }
+                 # Insure it's for one of our genomes.
+                 my $genomeID = FIG::genome_of($peg);
+                 if (exists $genomeHash->{$genomeID}) {
+                     # Verify the synonym.
+                     if ($syn_id ne $current_syn) {
+                         # It's new, so put it in the group table.
+                         $loadSynonymGroup->Put($syn_id);
+                         $current_syn = $syn_id;
+                     }
+                     # Connect the synonym to the peg.
+                     $loadIsSynonymGroupFor->Put($syn_id, $peg);
+                     # Count this feature.
+                     $featureCount++;
+                     if ($featureCount % 1000 == 0) {
+                         Trace("$featureCount features processed.") if T(3);
+                     }
+                 }
+             }
+             Trace("$entryCount rows produced $featureCount features.") if T(2);
+         }
+     }
+     # Finish the load.
+     my $retVal = $self->_FinishAll();
+     return $retVal;
+ }
+ =head3 LoadFamilyData
+     my $stats = $spl->LoadFamilyData();
+ Load the protein families into Sprout.
+ The following relations are loaded by this method.
+     Family
+     IsFamilyForFeature
+ The source information for these relations is taken from the C<families_for_protein>,
+ C<family_function>, and C<sz_family> methods of the B<FIG> object.
+ =over 4
+ =item RETURNS
+ Returns a statistics object for the loads.
+ =back
+ =cut
+ #: Return Type $%;
+ sub LoadFamilyData {
+     # Get this object instance.
+     my ($self) = @_;
+     # Get the FIG object.
+     my $fig = $self->{fig};
+     # Get the genome hash.
+     my $genomeHash = $self->{genomes};
+     # Create load objects for the tables we're loading.
+     my $loadFamily = $self->_TableLoader('Family');
+     my $loadIsFamilyForFeature = $self->_TableLoader('IsFamilyForFeature');
      if ($self->{options}->{loadOnly}) {
          Trace("Loading from existing files.") if T(2);
      } else {
-         Trace("Generating group data.") if T(2);
+         Trace("Generating family data.") if T(2);
+         # Create a hash for the family IDs.
+         my %familyHash = ();
          # Loop through the genomes.
-         my $line;
+         for my $genomeID (sort keys %{$genomeHash}) {
-         for my $genomeID (keys %{$genomeHash}) {
+             Trace("Processing features for $genomeID.") if T(2);
-             Trace("Processing $genomeID.") if T(3);
+             # Loop through this genome's PEGs.
-             # Open the NMPDR group file for this genome.
+             for my $fid ($fig->all_features($genomeID, "peg")) {
-             if (open(TMP, "<$FIG_Config::organisms/$genomeID/NMPDR") &&
+                 $loadIsFamilyForFeature->Add("features", 1);
-                 defined($line = <TMP>)) {
+                 # Get this feature's families.
-                 # Clean the line ending.
+                 my @families = $fig->families_for_protein($fid);
-                 chomp $line;
+                 # Loop through the families, connecting them to the feature.
-                 # Add the group to the table. Note that there can only be one group
+                 for my $family (@families) {
-                 # per genome.
+                     $loadIsFamilyForFeature->Put($family, $fid);
-                 $loadGenomeGroups->Put($genomeID, $line);
+                     # If this is a new family, create a record for it.
+                     if (! exists $familyHash{$family}) {
+                         $familyHash{$family} = 1;
+                         $loadFamily->Add("families", 1);
+                         my $size = $fig->sz_family($family);
+                         my $func = $fig->family_function($family);
+                         $loadFamily->Put($family, $size, $func);
+                     }
+                 }
              }
-             close TMP;
          }
      }
      # Finish the load.
-Line 1381
+Line 1785
      return $retVal;
  }
+ =head3 LoadDrugData
+     my $stats = $spl->LoadDrugData();
+ Load the drug target data into Sprout.
+ The following relations are loaded by this method.
+     PDB
+     DocksWith
+     IsProteinForFeature
+     Ligand
+ The source information for these relations is taken from attributes. The
+ C<PDB> attribute links a PDB to a feature, and is used to build B<IsProteinForFeature>.
+ The C<zinc_name> attribute describes the ligands. The C<docking_results>
+ attribute contains the information for the B<DocksWith> relationship. It is
+ expected that additional attributes and tables will be added in the future.
+ =over 4
+ =item RETURNS
+ Returns a statistics object for the loads.
+ =back
+ =cut
+ #: Return Type $%;
+ sub LoadDrugData {
+     # Get this object instance.
+     my ($self) = @_;
+     # Get the FIG object.
+     my $fig = $self->{fig};
+     # Get the genome hash.
+     my $genomeHash = $self->{genomes};
+     # Create load objects for the tables we're loading.
+     my $loadPDB = $self->_TableLoader('PDB');
+     my $loadLigand = $self->_TableLoader('Ligand');
+     my $loadIsProteinForFeature = $self->_TableLoader('IsProteinForFeature');
+     my $loadDocksWith = $self->_TableLoader('DocksWith');
+     if ($self->{options}->{loadOnly}) {
+         Trace("Loading from existing files.") if T(2);
+     } else {
+         Trace("Generating drug target data.") if T(2);
+         # First comes the "DocksWith" relationship. This will give us a list of PDBs.
+         # We can also encounter PDBs when we process "IsProteinForFeature". To manage
+         # this process, PDB information is collected in a hash table and then
+         # unspooled after both relationships are created.
+         my %pdbHash = ();
+         Trace("Generating docking data.") if T(2);
+         # Get all the docking data. This may cause problems if there are too many PDBs,
+         # at which point we'll need another algorithm. The indicator that this is
+         # happening will be a timeout error in the next statement.
+         my @dockData = $fig->query_attributes('$key = ? AND $value < ?',
+                                               ['docking_results', $FIG_Config::dockLimit]);
+         Trace(scalar(@dockData) . " rows of docking data found.") if T(3);
+         for my $dockData (@dockData) {
+             # Get the docking data components.
+             my ($pdbID, $docking_key, @valueData) = @{$dockData};
+             # Fix the PDB ID. It's supposed to be lower-case, but this does not always happen.
+             $pdbID = lc $pdbID;
+             # Strip off the object type.
+             $pdbID =~ s/pdb://;
+             # Extract the ZINC ID from the docking key. Note that there are two possible
+             # formats.
+             my (undef, $zinc_id) = $docking_key =~ /^docking_results::(ZINC)?(\d+)$/;
+             if (! $zinc_id) {
+                 Trace("Invalid docking result key $docking_key for $pdbID.") if T(0);
+                 $loadDocksWith->Add("errors");
+             } else {
+                 # Get the pieces of the value and parse the energy.
+                 # Note that we don't care about the rank, since
+                 # we can sort on the energy level itself in our database.
+                 my ($energy, $tool, $type) = @valueData;
+                 my ($rank, $total, $vanderwaals, $electrostatic) = split /\s*;\s*/, $energy;
+                 # Ignore predicted results.
+                 if ($type ne "Predicted") {
+                     # Count this docking result.
+                     if (! exists $pdbHash{$pdbID}) {
+                         $pdbHash{$pdbID} = 1;
+                     } else {
+                         $pdbHash{$pdbID}++;
+                     }
+                     # Write the result to the output.
+                     $loadDocksWith->Put($pdbID, $zinc_id, $electrostatic, $type, $tool,
+                                         $total, $vanderwaals);
+                 }
+             }
+         }
+         Trace("Connecting features.") if T(2);
+         # Loop through the genomes.
+         for my $genome (sort keys %{$genomeHash}) {
+             Trace("Generating PDBs for $genome.") if T(3);
+             # Get all of the PDBs that BLAST against this genome's features.
+             my @attributeData = $fig->get_attributes("fig|$genome%", 'PDB::%');
+             for my $pdbData (@attributeData) {
+                 # The PDB ID is coded as a subkey.
+                 if ($pdbData->[1] !~ /PDB::(.+)/i) {
+                     Trace("Invalid PDB ID \"$pdbData->[1]\" in attribute table.") if T(0);
+                     $loadPDB->Add("errors");
+                 } else {
+                     my $pdbID = $1;
+                     # Insure the PDB is in the hash.
+                     if (! exists $pdbHash{$pdbID}) {
+                         $pdbHash{$pdbID} = 0;
+                     }
+                     # The score and locations are coded in the attribute value.
+                     if ($pdbData->[2] !~ /^([^;]+)(.*)$/) {
+                         Trace("Invalid PDB data for $pdbID and feature $pdbData->[0].") if T(0);
+                         $loadIsProteinForFeature->Add("errors");
+                     } else {
+                         my ($score, $locData) = ($1,$2);
+                         # The location data may not be present, so we have to start with some
+                         # defaults and then check.
+                         my ($start, $end) = (1, 0);
+                         if ($locData) {
+                             $locData =~ /(\d+)-(\d+)/;
+                             $start = $1;
+                             $end = $2;
+                         }
+                         # If we still don't have the end location, compute it from
+                         # the feature length.
+                         if (! $end) {
+                             # Most features have one location, but we do a list iteration
+                             # just in case.
+                             my @locations = $fig->feature_location($pdbData->[0]);
+                             $end = 0;
+                             for my $loc (@locations) {
+                                 my $locObject = BasicLocation->new($loc);
+                                 $end += $locObject->Length;
+                             }
+                         }
+                         # Decode the score.
+                         my $realScore = FIGRules::DecodeScore($score);
+                         # Connect the PDB to the feature.
+                         $loadIsProteinForFeature->Put($pdbID, $pdbData->[0], $start, $realScore, $end);
+                     }
+                 }
+             }
+         }
+         # We've got all our PDBs now, so we unspool them from the hash.
+         Trace("Generating PDBs. " . scalar(keys %pdbHash) . " found.") if T(2);
+         my $count = 0;
+         for my $pdbID (sort keys %pdbHash) {
+             $loadPDB->Put($pdbID, $pdbHash{$pdbID});
+             $count++;
+             Trace("$count PDBs processed.") if T(3) && ($count % 500 == 0);
+         }
+         # Finally we create the ligand table. This information can be found in the
+         # zinc_name attribute.
+         Trace("Loading ligands.") if T(2);
+         # The ligand list is huge, so we have to get it in pieces. We also have to check for duplicates.
+         my $last_zinc_id = "";
+         my $zinc_id = "";
+         my $done = 0;
+         while (! $done) {
+             # Get the next 10000 ligands. We insist that the object ID is greater than
+             # the last ID we processed.
+             Trace("Loading batch starting with ZINC:$zinc_id.") if T(3);
+             my @attributeData = $fig->query_attributes('$object > ? AND $key = ? ORDER BY $object LIMIT 10000',
+                                                        ["ZINC:$zinc_id", "zinc_name"]);
+             Trace(scalar(@attributeData) . " attribute rows returned.") if T(3);
+             if (! @attributeData) {
+                 # Here there are no attributes left, so we quit the loop.
+                 $done = 1;
+             } else {
+                 # Process the attribute data we've received.
+                 for my $zinc_data (@attributeData) {
+                     # The ZINC ID is found in the first return column, prefixed with the word ZINC.
+                     if ($zinc_data->[0] =~ /^ZINC:(\d+)$/) {
+                         $zinc_id = $1;
+                         # Check for a duplicate.
+                         if ($zinc_id eq $last_zinc_id) {
+                             $loadLigand->Add("duplicate");
+                         } else {
+                             # Here it's safe to output the ligand. The ligand name is the attribute value
+                             # (third column in the row).
+                             $loadLigand->Put($zinc_id, $zinc_data->[2]);
+                             # Insure we don't try to add this ID again.
+                             $last_zinc_id = $zinc_id;
+                         }
+                     } else {
+                         Trace("Invalid zinc ID \"$zinc_data->[0]\" in attribute table.") if T(0);
+                         $loadLigand->Add("errors");
+                     }
+                 }
+             }
+         }
+         Trace("Ligands loaded.") if T(2);
+     }
+     # Finish the load.
+     my $retVal = $self->_FinishAll();
+     return $retVal;
+ }
  =head2 Internal Utility Methods
+ =head3 SpecialAttribute
+     my $count = SproutLoad::SpecialAttribute($id, \@attributes, $idxMatch, \@idxValues, $pattern, $loader);
+ Look for special attributes of a given type. A special attribute is found by comparing one of
+ the columns of the incoming attribute list to a search pattern. If a match is found, then
+ a set of columns is put into an output table connected to the specified ID.
+ For example, when processing features, the attribute list we look at has three columns: attribute
+ name, attribute value, and attribute value HTML. The IEDB attribute exists if the attribute name
+ begins with C<iedb_>. The call signature is therefore
+     my $found = SpecialAttribute($fid, \@attributeList, 0, [0,2], '^iedb_', $loadFeatureIEDB);
+ The pattern is matched against column 0, and if we have a match, then column 2's value is put
+ to the output along with the specified feature ID.
+ =over 4
+ =item id
+ ID of the object whose special attributes are being loaded. This forms the first column of the
+ output.
+ =item attributes
+ Reference to a list of tuples.
+ =item idxMatch
+ Index in each tuple of the column to be matched against the pattern. If the match is
+ successful, an output record will be generated.
+ =item idxValues
+ Reference to a list containing the indexes in each tuple of the columns to be put as
+ the second column of the output.
+ =item pattern
+ Pattern to be matched against the specified column. The match will be case-insensitive.
+ =item loader
+ An object to which each output record will be put. Usually this is an B<ERDBLoad> object,
+ but technically it could be anything with a C<Put> method.
+ =item RETURN
+ Returns a count of the matches found.
+ =item
+ =back
+ =cut
+ sub SpecialAttribute {
+     # Get the parameters.
+     my ($id, $attributes, $idxMatch, $idxValues, $pattern, $loader) = @_;
+     # Declare the return variable.
+     my $retVal = 0;
+     # Loop through the attribute rows.
+     for my $row (@{$attributes}) {
+         # Check for a match.
+         if ($row->[$idxMatch] =~ m/$pattern/i) {
+             # We have a match, so output a row. This is a bit tricky, since we may
+             # be putting out multiple columns of data from the input.
+             my $value = join(" ", map { $row->[$_] } @{$idxValues});
+             $loader->Put($id, $value);
+             $retVal++;
+         }
+     }
+     Trace("$retVal special attributes found for $id and loader " . $loader->RelName() . ".") if T(4) && $retVal;
+     # Return the number of matches.
+     return $retVal;
+ }
  =head3 TableLoader
  Create an ERDBLoad object for the specified table. The object is also added to
-Line 1397
+Line 2076
  Name of the table (relation) being loaded.
- =item ignore
- TRUE if the table should be ignored entirely, else FALSE.
  =item RETURN
  Returns an ERDBLoad object for loading the specified table.
-Line 1411
+Line 2086
  sub _TableLoader {
      # Get the parameters.
-     my ($self, $tableName, $ignore) = @_;
+     my ($self, $tableName) = @_;
      # Create the load object.
-     my $retVal = ERDBLoad->new($self->{erdb}, $tableName, $self->{loadDirectory}, $self->LoadOnly,
+     my $retVal = ERDBLoad->new($self->{erdb}, $tableName, $self->{loadDirectory}, $self->LoadOnly);
-                                $ignore);
      # Cache it in the loader list.
      push @{$self->{loaders}}, $retVal;
      # Return it to the caller.
-Line 1448
+Line 2122
      my $retVal = Stats->new();
      # Get the loader list.
      my $loadList = $self->{loaders};
+     # Create a hash to hold the statistics objects, keyed on relation name.
+     my %loaderHash = ();
      # Loop through the list, finishing the loads. Note that if the finish fails, we die
-     # ignominiously. At some future point, we want to make the loads restartable.
+     # ignominiously. At some future point, we want to make the loads more restartable.
      while (my $loader = pop @{$loadList}) {
          # Get the relation name.
          my $relName = $loader->RelName;
-Line 1460
+Line 2136
              # Here we really need to finish.
              Trace("Finishing $relName.") if T(2);
              my $stats = $loader->Finish();
-             if ($self->{options}->{dbLoad} && ! $loader->Ignore) {
+             $loaderHash{$relName} = $stats;
+         }
+     }
+     # Now we loop through again, actually loading the tables. We want to finish before
+     # loading so that if something goes wrong at this point, all the load files are usable
+     # and we don't have to redo all that work.
+     for my $relName (sort keys %loaderHash) {
+         # Get the statistics for this relation.
+         my $stats = $loaderHash{$relName};
+         # Check for a database load.
+         if ($self->{options}->{dbLoad}) {
                  # Here we want to use the load file just created to load the database.
                  Trace("Loading relation $relName.") if T(2);
                  my $newStats = $self->{sprout}->LoadUpdate(1, [$relName]);
-Line 1470
+Line 2156
              $retVal->Accumulate($stats);
              Trace("Statistics for $relName:\n" . $stats->Show()) if T(2);
          }
-     }
      # Return the load statistics.
      return $retVal;
  }
+ =head3 GetGenomeAttributes
+     my $aHashRef = GetGenomeAttributes($fig, $genomeID, \@fids, \@propKeys);
+ Return a hash of attributes keyed on feature ID. This method gets all the NMPDR-related
+ attributes for all the features of a genome in a single call, then organizes them into
+ a hash.
+ =over 4
+ =item fig
+ FIG-like object for accessing attributes.
+ =item genomeID
+ ID of the genome who's attributes are desired.
+ =item fids
+ Reference to a list of the feature IDs whose attributes are to be kept.
+ =item propKeys
+ A list of the keys to retrieve.
+ =item RETURN
+ Returns a reference to a hash. The key of the hash is the feature ID. The value is the
+ reference to a list of the feature's attribute tuples. Each tuple contains the feature ID,
+ the attribute key, and one or more attribute values.
+ =back
+ =cut
+ sub GetGenomeAttributes {
+     # Get the parameters.
+     my ($fig, $genomeID, $fids, $propKeys) = @_;
+     # Declare the return variable.
+     my $retVal = {};
+     # Initialize the hash. This not only enables us to easily determine which FIDs to
+     # keep, it insures that the caller sees a list reference for every known fid,
+     # simplifying the logic.
+     for my $fid (@{$fids}) {
+         $retVal->{$fid} = [];
+     }
+     # Get the attributes. If ev_code_cron is running, we may get a timeout error, so
+     # an eval is used.
+     my @aList = ();
+     eval {
+         @aList = $fig->get_attributes("fig|$genomeID%", $propKeys);
+         Trace(scalar(@aList) . " attributes returned for genome $genomeID.") if T(3);
+     };
+     # Check for a problem.
+     if ($@) {
+         Trace("Retrying attributes for $genomeID due to error: $@") if T(1);
+         # Our fallback plan is to process the attributes in blocks of 100. This is much slower,
+         # but allows us to continue processing.
+         my $nFids = scalar @{$fids};
+         for (my $i = 0; $i < $nFids; $i += 100) {
+             # Determine the index of the last feature ID we'll be specifying on this pass.
+             # Normally it's $i + 99, but if we're close to the end it may be less.
+             my $end = ($i + 100 > $nFids ? $nFids - 1 : $i + 99);
+             # Get a slice of the fid list.
+             my @slice = @{$fids}[$i .. $end];
+             # Get the relevant attributes.
+             Trace("Retrieving attributes for fids $i to $end.") if T(3);
+             my @aShort = $fig->get_attributes(\@slice, $propKeys);
+             Trace(scalar(@aShort) . " attributes returned for fids $i to $end.") if T(3);
+             push @aList, @aShort;
+         }
+     }
+     # Now we should have all the interesting attributes in @aList. Populate the hash with
+     # them.
+     for my $aListEntry (@aList) {
+         my $fid = $aListEntry->[0];
+         if (exists $retVal->{$fid}) {
+             push @{$retVal->{$fid}}, $aListEntry;
+         }
+     }
+     # Return the result.
+     return $retVal;
+ }
 ;

 Legend:



Removed from v.1.26
 


changed lines


 
Added in v.1.94
 Legend:



Removed from v.1.26
 


changed lines


 
Added in v.1.94
-Removed from v.1.26
+Added in v.1.94

MCS Webmaster	ViewVC Help
Powered by ViewVC 1.0.3