Initial setup of new biodata backend for elasticsearch

biodata-elastic-setup.diff

# git diff
diff --git a/ga4gh/datamodel/datasets.py b/ga4gh/datamodel/datasets.py
index 25f3f7f..c07537e 100644
--- a/ga4gh/datamodel/datasets.py
+++ b/ga4gh/datamodel/datasets.py
@@ -358,3 +358,66 @@ class FileSystemDataset(AbstractDataset):
                 except KeyError as err:
                     raise exceptions.MissingDatasetMetadataException(
                         metadataFileName, str(err))
+
+class FileSystemElasticSearchDataset(FileSystemDataset):
+    """
+    Similar to FileSystemDataset, except the biodata is served from
+    elasticsearch.
+    """
+
+    def __init__(self, localId, dataDir, dataRepository):
+        super(FileSystemElasticSearchDataset, self).__init__(localId)
+
+    def addBioSample(self, bioSample):
+        id_ = bioSample.getId()
+        # TODO - call ES
+        pass
+
+    def addIndividual(self, individual):
+        id_ = individual.getId()
+        # TODO - call ES
+        pass
+
+    def getBioSample(self, id_):
+        """
+        Returns the BioSample with the specified ID, or raises a
+        BioSampleNotFoundException otherwise.
+        """
+        # TODO - call ES
+        pass
+
+    def getBioSamples(self):
+        """
+        Returns all BioSamples in this dataset
+        """
+        # TODO - call ES
+        pass
+
+    def getIndividual(self, id_):
+        """
+        Returns the Individual with the specified ID, or raises a
+        IndividualNotFoundException otherwise.
+        """
+        # TODO - call ES
+        pass
+
+    def getIndividuals(self):
+        """
+        Returns all Individuals in this dataset
+        """
+        # TODO - call ES
+        pass
+
+    def getBioSampleByIndex(self, index):
+        """
+        Returns the BioSample set at the specified index in this dataset.
+        """
+        # TODO - call ES
+        pass
+
+    def getIndividualByIndex(self, index):
+        """
+        Returns the Individual set at the specified index in this dataset.
+        """
+        # TODO - call ES
+        pass
diff --git a/ga4gh/datarepo.py b/ga4gh/datarepo.py
index 14a265c..9571e2c 100644
--- a/ga4gh/datarepo.py
+++ b/ga4gh/datarepo.py
@@ -212,3 +212,11 @@ class FileSystemDataRepository(AbstractDataRepository):
                 readGroupSet.checkConsistency(self)
             for variantSet in dataset.getVariantSets():
                 variantSet.checkConsistency()
+
+class FileSystemElasticSearchDataRepository(FileSystemDataRepository):
+
+    def __init__(self, dataDir):
+        super(FileSystemElasticSearchDataRepository, self).__init__()
+        constructors = [
+            references.HtslibReferenceSet, ontologies.FileSystemOntologies,
+            datasets.FileSystemElasticSearchDataset]
diff --git a/ga4gh/frontend.py b/ga4gh/frontend.py
index 849e154..84d6257 100644
# ideally adding a backend shouldn't mean having to change so many (or any) existing core python file

--- a/ga4gh/frontend.py
+++ b/ga4gh/frontend.py
@@ -213,6 +213,10 @@ def configure(configFile=None, baseConfig="ProductionConfig",
     elif dataSource.scheme == "file":
         dataRepository = datarepo.FileSystemDataRepository(os.path.join(
             dataSource.netloc, dataSource.path))
+    elif dataSource.scheme == "file-elasticsearch":
+        dataRepository = datarepo.FileSystemElasticSearchDataRepository(
+            os.path.join(dataSource.netloc, dataSource.path))
+
     else:
         raise exceptions.ConfigurationException(
             "Unsupported data source scheme: " + dataSource.scheme)