Make referencesets optional for variant sets

Allow nullity for referencesetid field in models
Safely access the referenceset name in a few places

ga4gh/server/cli/repomanager.py

@@ -275,16 +275,22 @@ def addVariantSet(self):
         variantSet.populateFromFile(dataUrls, indexFiles)
         # Get the reference set that is associated with the variant set.
         referenceSetName = self._args.referenceSetName
-        if referenceSetName is None:
-            # Try to find a reference set name from the VCF header.
-            referenceSetName = variantSet.getVcfHeaderReferenceSetName()
-        if referenceSetName is None:
-            raise exceptions.RepoManagerException(
-                "Cannot infer the ReferenceSet from the VCF header. Please "
-                "specify the ReferenceSet to associate with this "
-                "VariantSet using the --referenceSetName option")
-        referenceSet = self._repo.getReferenceSetByName(referenceSetName)
-        variantSet.setReferenceSet(referenceSet)
+        referenceSet = None
+        if referenceSetName is not None:
+            referenceSet = self._repo.getReferenceSetByName(referenceSetName)
+        elif variantSet.getVcfHeaderReferenceSetName() is not None:
+            try:
+                referenceSet = self._repo.getReferenceSetByName(
+                    variantSet.getVcfHeaderReferenceSetName())
+            except exceptions.ReferenceSetNameNotFoundException as e:
+                # It's a lucky shot to get a reference set by name using the
+                # VCF header.
+                pass
+        print('made it')
+        if referenceSet is not None:
+            print('hur')
+
+            variantSet.setReferenceSet(referenceSet)
         variantSet.setAttributes(json.loads(self._args.attributes))
         # Now check for annotations
         annotationSets = []

ga4gh/server/datarepo.py

@@ -221,9 +221,13 @@ def printSummary(self):
                         sep="\t")
             print("\tVariantSets:")
             for variantSet in dataset.getVariantSets():
+                referenceSetName = ""
+                if variantSet.getReferenceSet() is not None:
+                    referenceSetName = \
+                        variantSet.getReferenceSet().getLocalId()
                 print(
                     "\t", variantSet.getLocalId(),
-                    variantSet.getReferenceSet().getLocalId(),
+                    referenceSetName,
                     variantSet.getId(),
                     sep="\t")
                 if variantSet.getNumVariantAnnotationSets() > 0:
@@ -235,17 +239,25 @@ def printSummary(self):
                             vas.getOntology().getName(), sep="\t")
             print("\tFeatureSets:")
             for featureSet in dataset.getFeatureSets():
+                referenceSetName = ""
+                if featureSet.getReferenceSet() is not None:
+                    referenceSetName = \
+                        featureSet.getReferenceSet().getLocalId()
                 print(
                     "\t", featureSet.getLocalId(),
-                    featureSet.getReferenceSet().getLocalId(),
+                    referenceSetName,
                     featureSet.getOntology().getName(),
                     featureSet.getId(),
                     sep="\t")
             print("\tContinuousSets:")
             for continuousSet in dataset.getContinuousSets():
+                referenceSetName = ""
+                if continuousSet.getReferenceSet() is not None:
+                    referenceSetName = \
+                        continuousSet.getReferenceSet().getLocalId()
                 print(
                     "\t", continuousSet.getLocalId(),
-                    continuousSet.getReferenceSet().getLocalId(),
+                    referenceSetName,
                     continuousSet.getId(),
                     sep="\t")
             print("\tPhenotypeAssociationSets:")
@@ -1016,11 +1028,14 @@ def insertVariantSet(self, variantSet):
             [protocol.toJsonDict(metadata) for metadata in
              variantSet.getMetadata()])
         urlMapJson = json.dumps(variantSet.getReferenceToDataUrlIndexMap())
+        referenceSetId = None
+        if variantSet.getReferenceSet() is not None:
+            referenceSetId = variantSet.getReferenceSet().getId()
         try:
             m.Variantset.create(
                 id=variantSet.getId(),
                 datasetid=variantSet.getParentContainer().getId(),
-                referencesetid=variantSet.getReferenceSet().getId(),
+                referencesetid=referenceSetId,
                 name=variantSet.getLocalId(),
                 created=datetime.datetime.now(),
                 updated=datetime.datetime.now(),
@@ -1035,11 +1050,13 @@ def insertVariantSet(self, variantSet):
     def _readVariantSetTable(self):
         for variantSetRecord in m.Variantset.select():
             dataset = self.getDataset(variantSetRecord.datasetid.id)
-            referenceSet = self.getReferenceSet(
-                variantSetRecord.referencesetid.id)
             variantSet = variants.HtslibVariantSet(
                 dataset, variantSetRecord.name)
-            variantSet.setReferenceSet(referenceSet)
+            if variantSetRecord.referencesetid is not None:
+                referenceSet = self.getReferenceSet(
+                    variantSetRecord.referencesetid.id)
+                variantSet.setReferenceSet(referenceSet)
+
             variantSet.populateFromRow(variantSetRecord)
             assert variantSet.getId() == variantSetRecord.id
             # Insert the variantSet into the memory-based object model.

ga4gh/server/repo/models.py

@@ -94,7 +94,7 @@ class Variantset(BaseModel):
     metadata = pw.TextField(null=True)
     name = pw.TextField()
     referencesetid = pw.ForeignKeyField(
-        db_column='referenceSetId', rel_model=Referenceset, to_field='id')
+        db_column='referenceSetId', rel_model=Referenceset, to_field='id', null=True)
     updated = pw.TextField(null=True)
 
     class Meta:
@@ -138,7 +138,7 @@ class Featureset(BaseModel):
     ontologyid = pw.ForeignKeyField(
         db_column='ontologyId', rel_model=Ontology, to_field='id')
     referencesetid = pw.ForeignKeyField(
-        db_column='referenceSetId', rel_model=Referenceset, to_field='id')
+        db_column='referenceSetId', rel_model=Referenceset, to_field='id', null=True)
     sourceuri = pw.TextField(
         db_column='sourceUri', null=True)
 
@@ -157,7 +157,7 @@ class ContinuousSet(BaseModel):
     info = pw.TextField(null=True)
     name = pw.TextField()
     referencesetid = pw.ForeignKeyField(
-        db_column='referenceSetId', rel_model=Referenceset, to_field='id')
+        db_column='referenceSetId', rel_model=Referenceset, to_field='id', null=True)
     sourceuri = pw.TextField(
         db_column='sourceUri', null=True)
 
@@ -270,7 +270,7 @@ class Rnaquantificationset(BaseModel):
     info = pw.TextField(null=True)
     name = pw.TextField()
     referencesetid = pw.ForeignKeyField(
-        db_column='referenceSetId', rel_model=Referenceset, to_field='id')
+        db_column='referenceSetId', rel_model=Referenceset, to_field='id', null=True)
 
     class Meta:
         db_table = 'RnaQuantificationSet'