All of Bob's modifications since September

ISA/lib/perl/Commentable.pm

@@ -28,7 +28,7 @@ sub new {
     my $commentValues = $args->{$ck};
     next unless $commentValues;
 
-    $ck =~ m/$commentString\[(.+)\]/;
+    $ck =~ m/${commentString}_?\[(.+)\]/;
     my $comment = CBIL::ISA::StudyAssayEntity::Comment->new({'_qualifier' => $1, '_value' => $commentValues});
     push @{$obj->{_comments}}, $comment;
   }

ISA/lib/perl/Investigation.pm

@@ -26,6 +26,7 @@ my $STUDY_PROTOCOLS = "STUDY PROTOCOLS";
 my $STUDY_FACTORS = "STUDY FACTORS";
 my $STUDY_PUBLICATIONS = "STUDY PUBLICATIONS";
 my $STUDY_ASSAYS = "STUDY ASSAYS";
+my $STUDY_TAGS = "STUDY TAGS"; # extension introduced by VectorBase
 
 
 sub setDebug {$_[0]->{_debug} = $_[1]}
@@ -110,6 +111,9 @@ sub makeStudy {
   $study->setProtocols($hash->{$STUDY_PROTOCOLS}, 
                          $columnCounts->{$STUDY_PROTOCOLS});
 
+  $study->setTags($hash->{$STUDY_TAGS}, 
+                            $columnCounts->{$STUDY_TAGS});
+
   return $study;
 }
 
@@ -186,6 +190,22 @@ sub parseStudy {
     $assayFileReader->closeFh();
   }
 
+  foreach my $file (@{$study->getAssayDataFiles()}) {
+    # only process phenotype (p_*) and genotype (g_*) ISA-like files
+    if ($file->getValue() =~ /^(?:p_|g_)/) {
+      my $assayDataFileName = $investigationDirectory . "/" . $file->getValue();
+      warn "About to load $assayDataFileName...\n";
+
+      my $assayDataFileReader = CBIL::ISA::StudyAssayFileReader->new($assayDataFileName, $delimiter);
+
+      while($assayDataFileReader->hasNextLine()) {
+        my $dataObjects = $assayDataFileReader->readLineToObjects();
+        $study->addNodesAndEdges($dataObjects, $assayDataFileName);
+      }
+      $assayDataFileReader->closeFh();
+    }
+  }
+
   $study->setHasMoreData(0);
 }
 
@@ -251,8 +271,8 @@ sub dealWithAllOntologies {
     my $term = $ontologyTerm->getTerm();
 
 
-    unless(($accession && $source) || blessed($ontologyTerm) eq 'CBIL::ISA::StudyAssayEntity::Characteristic' || blessed($ontologyTerm) eq 'CBIL::ISA::StudyAssayEntity::ParameterValue') {
-      $self->handleError("OntologyTerm $term is required to have accession and source.");
+    unless(($accession && $source) or not $ontologyTerm->requiresAccessionedTerm()) {
+      $self->handleError("OntologyTerm '$term' (context: ".$ontologyTerm->getDebugContext().") is required to have accession and source.");
     }
   }
 }

ISA/lib/perl/InvestigationFileReader.pm

@@ -30,6 +30,7 @@ my @CONTEXTS = ("ONTOLOGY SOURCE REFERENCE",
                 "STUDY ASSAYS",
                 "STUDY PROTOCOLS",
                 "STUDY CONTACTS",
+                "STUDY TAGS", # extension by VectorBase
     );
 
 sub new {
@@ -63,6 +64,9 @@ sub read {
     # split and remove leading and trailing quotes
     my @a = map { s/^"(.*)"$/$1/; $_; } split($delimiter, $_);
 
+    # skip completely empty lines
+    next unless (length(join '', @a)>0);
+
     $studyIdentifier = $a[1] if(uc $a[0] eq 'STUDY IDENTIFIER');
     if(&isContextSwitch($a[0])) {
       $lineContext = uc $a[0];

ISA/lib/perl/OntologyTerm.pm

@@ -18,13 +18,25 @@ sub getTermAccessionNumber { $_[0]->{_term_accession_number} }
 sub setTermSourceRef { $_[0]->{_term_source_ref} = $_[1] }
 sub getTermSourceRef { $_[0]->{_term_source_ref} }
 
+=head2 getDebugContext setDebugContext
+
+Getter/setter for a string that gives an idea where the ontology term was attempted to be loaded from.
+This can help debugging poorly constructed ISA-Tab.
+
+=cut
+
+sub setDebugContext { $_[0]->{_debug_context} = $_[1] }
+sub getDebugContext { $_[0]->{_debug_context} }
+
 sub new {
   my ($class, $args) = @_;
 
   my $seenTerm;
 
   my $self = bless {}, $class;
 
+  $self->setDebugContext(delete $args->{_debug_context}) if ($args->{_debug_context});
+
   foreach my $key (keys %$args) {
     if($key =~ /term_source_ref/) {
       $self->setTermSourceRef($args->{$key});
@@ -58,4 +70,15 @@ sub getValue { $_[0]->getTerm() }
 sub setValue { $_[0]->setTerm($_[1]) }
 
 
+#
+# if this column must contain a fully specified ontology term
+# then return true
+#
+# e.g. Characteristics and Parameter Value columns will override this to return false,
+# because they can contain free text and/or numbers+units
+#
+sub requiresAccessionedTerm {
+  return 1;
+}
+
 1;

ISA/lib/perl/Study.pm

@@ -7,6 +7,7 @@ use CBIL::ISA::Publication;
 use CBIL::ISA::Contact;
 use CBIL::ISA::OntologyTerm;
 use CBIL::ISA::StudyDesign;
+use CBIL::ISA::Tag;
 use CBIL::ISA::StudyFactor;
 use CBIL::ISA::StudyAssay;
 use CBIL::ISA::Protocol;
@@ -91,7 +92,21 @@ sub setStudyDesigns {
   return $self->getStudyDesigns();
 }
 sub getStudyDesigns { $_[0]->{_study_designs}  || [] }
-sub addStudyDesign { push @{$_[0]->{_study_design}}, $_[1] }
+sub addStudyDesign { push @{$_[0]->{_study_designs}}, $_[1] }
+
+
+sub setTags { 
+  my ($self, $hash, $columnCount) = @_;
+
+  my $otRegexs = ["_tag"];
+  my $otIsList = [ 0 ];
+  $self->{_tags} = $self->makeStudyObjectsFromHash($hash, $columnCount, "CBIL::ISA::Tag", $otRegexs, $otIsList);
+
+  return $self->getTags();
+}
+sub getTags { $_[0]->{_tags}  || [] }
+sub addTag { push @{$_[0]->{_tags}}, $_[1] }
+
 
 sub setStudyFactors { 
   my ($self, $hash, $columnCount) = @_;
@@ -169,6 +184,26 @@ sub addEdge {
 }
 sub getEdges { $_[0]->{_edges} or [] }
 
+#
+# used to remember which File entities, such as "Raw Data File"
+# have been encountered while parsing assay files
+#
+sub getAssayDataFiles { $_[0]->{_assay_data_files} or [] }
+sub addAssayDataFile {
+  my ($self, $fileObj) = @_;
+
+  # check it's not already there
+  # can't use "$obj->equals($other)" because attributes
+  # are not nodes and are "never equal"
+  # so we just use the value (the filename)
+  foreach (@{$self->getAssayDataFiles()}) {
+    return $_ if ($fileObj->getValue() eq $_->getValue());
+  }
+  push @{$self->{_assay_data_files}}, $fileObj;
+  return $fileObj;
+}
+
+
 
 # Handle a chunk of the Investigation File
 #  Each section is made into an object
@@ -210,7 +245,6 @@ sub makeStudyObjectsFromHash {
       my $setterName = $setOrAdd . join("", map { ucfirst } split("_", $otRegex));
 
       my %initOtHash = map { $_ => $hash->{$_}->[$i] } @{$otKeys{$otRegex}};
-
       my %otHash;
 
       if($otIsList) {
@@ -226,13 +260,17 @@ sub makeStudyObjectsFromHash {
       }
 
       foreach my $n (keys %otHash) {
-        my $ontologyTerm = CBIL::ISA::OntologyTerm->new($otHash{$n});
-        eval {
-          $obj->$setterName($ontologyTerm);
-
-        };
-        if ($@) {
-          die "Unable to $setterName for class $class: $@";
+        my $otHash = $otHash{$n};
+        # don't create and attach a term if all values (name, accession, source) are empty.
+        if (length(join '', values %$otHash)>0) {
+          my $ontologyTerm = CBIL::ISA::OntologyTerm->new($otHash);
+          # $ontologyTerm->setDebugContext(join ', ',  map { "$_=$hash{$_}" } keys %hash);
+          eval {
+            $obj->$setterName($ontologyTerm);
+          };
+          if ($@) {
+            die "Unable to $setterName for class $class: $@";
+          }
         }
       }
     }
@@ -279,6 +317,10 @@ sub addNodesAndEdges {
 
     $start = 1;
     $wasNodeContext = $entity->isNode();
+
+    # now keep a track of all File entities
+    # for later processing by PopBio Genotype/Phenotype handler
+    map { $self->addAssayDataFile($_) } @{$entity->getFiles()} if ($entity->can('getFiles'));
   }
 }
 
@@ -374,6 +416,14 @@ inherited from L<CBIL::ISA::Commentable>
 
 @return array of L<CBIL::ISA::StudyDesign>
 
+=item C<addTag>
+
+@param L<CBIL::ISA::Tag>
+
+=item C<getTags>
+
+@return array of L<CBIL::ISA::Tag>
+
 =item C<addStudyFactor>
 
 @param L<CBIL::ISA::StudyFactors>

ISA/lib/perl/StudyAssayEntity/Assay.pm

@@ -1,6 +1,22 @@
 package CBIL::ISA::StudyAssayEntity::Assay;
-use base qw(CBIL::ISA::StudyAssayEntity::Fileable);
+use base qw(CBIL::ISA::StudyAssayEntity::DataEntity CBIL::ISA::StudyAssayEntity::Fileable);
 
 use strict;
 
+
+sub getAttributeNames {
+  my ($self) = @_;
+
+  my @attributeQualifiers = ();
+
+  my $attrs1 = $self->CBIL::ISA::StudyAssayEntity::DataEntity::getAttributeNames();
+  my $attrs2 = $self->CBIL::ISA::StudyAssayEntity::Fileable::getAttributeNames();
+  my $attrs = [ @$attrs1, @$attrs2 ];
+
+  push @{$attrs}, @attributeQualifiers;
+
+  return $attrs;
+}
+
+
 1;

ISA/lib/perl/StudyAssayEntity/Attribute.pm

@@ -0,0 +1,10 @@
+package CBIL::ISA::StudyAssayEntity::Attribute;
+use base qw(CBIL::ISA::OntologyTerm);
+
+use strict;
+
+sub getParents {
+  return [ "Phenotype" ];
+}
+
+1;

ISA/lib/perl/StudyAssayEntity/Characteristic.pm

@@ -4,7 +4,7 @@ use base qw(CBIL::ISA::OntologyTermWithQualifier);
 use strict;
 
 sub getParents {
-  return ["Source", "Sample", "Extract", "LabeledExtract" ];
+  return ["Source", "Sample", "Extract", "LabeledExtract", "Assay", "Phenotype", "Genotype" ];
 }
 
 # @override
@@ -52,4 +52,8 @@ sub qualifierContextMethod {
   return "addCharacteristic";
 }
 
+sub requiresAccessionedTerm {
+  return 0;
+}
+
 1;

ISA/lib/perl/StudyAssayEntity/DataEntity.pm

@@ -0,0 +1,32 @@
+package CBIL::ISA::StudyAssayEntity::DataEntity;
+use base qw(CBIL::ISA::StudyAssayEntity::Fileable);
+
+=head1
+
+This class has been added to allow Description and Characteristics
+attributes (as used in VectorBase PopBio) to be attached to "Assay",
+"Phenotype" and "Genotype" entities.
+
+=cut
+
+use strict;
+
+sub setDescription { $_[0]->{_description} = $_[1] }
+sub getDescription { $_[0]->{_description} }
+
+sub addCharacteristic { push @{$_[0]->{_characteristics}}, $_[1] }
+sub getCharacteristics { $_[0]->{_characteristics}  || [] }
+
+sub getAttributeNames {
+  my ($self) = @_;
+
+  my @attributeQualifiers = ("Characteristic", "Description");
+
+  my $attrs = $self->SUPER::getAttributeNames();
+
+  push @{$attrs}, @attributeQualifiers;
+
+  return $attrs;
+}
+
+1;

ISA/lib/perl/StudyAssayEntity/Description.pm

@@ -10,7 +10,7 @@ sub getAttributeNames {
 }
 
 sub getParents {
-  return ["Source", "Sample", "Extract", "LabeledExtract"];
+  return ["Source", "Sample", "Extract", "LabeledExtract", "Assay"];
 }
 
 

ISA/lib/perl/StudyAssayEntity/Genotype.pm

@@ -0,0 +1,22 @@
+package CBIL::ISA::StudyAssayEntity::Genotype;
+use base qw(CBIL::ISA::StudyAssayEntity::DataEntity);
+
+use strict;
+
+sub setType { $_[0]->{_genotype_type} = $_[1] }
+sub getType { $_[0]->{_genotype_type} }
+
+sub getAttributeNames {
+  my ($self) = @_;
+
+  my @attributeQualifiers = ("Type");
+
+  my $attrs = $self->SUPER::getAttributeNames();
+
+  push @{$attrs}, @attributeQualifiers;
+
+  return $attrs;
+}
+
+
+1;

ISA/lib/perl/StudyAssayEntity/Observable.pm

@@ -0,0 +1,10 @@
+package CBIL::ISA::StudyAssayEntity::Observable;
+use base qw(CBIL::ISA::OntologyTerm);
+
+use strict;
+
+sub getParents {
+  return [ "Phenotype" ];
+}
+
+1;

ISA/lib/perl/StudyAssayEntity/ParameterValue.pm

@@ -31,4 +31,8 @@ sub qualifierContextMethod {
   return "addParameterValue";
 }
 
+sub requiresAccessionedTerm {
+  return 0;
+}
+
 1;