Merge pull request #16 from GenomicMedLab/issue-13

feat #13: get statistics about variation content

.gitignore

@@ -12,6 +12,7 @@ build/
 develop-eggs/
 dist/
 downloads/
+*.DS_Store
 eggs/
 .eggs/
 .idea/

src/anyvar/restapi/_data/openapi.yaml

@@ -433,6 +433,34 @@ paths:
                 type: object
                 # This seems to be working: $ref: "file:vr.json#/definitions/Location"
 
+  #######################################################################
+  # Summary Statistics
+
+  /summary_statistics/{vartype}:
+    get:
+      tags: ["Statistics"]
+      summary: Summary Statistics for Variations in anyvar
+      description: Output summary statistics
+      parameters:
+        - name: vartype
+          in: path
+          description: variation type
+          required: true
+          schema:
+            type: string
+            default: all
+            enum:
+              - substitution
+              - deletion
+              - insertion
+              - all
+      responses:
+        200:
+          description: OK
+          content:
+            application/json:
+              schema:
+                type: integer
 
 
 # ######################################################################

src/anyvar/restapi/routes/summary_statistics.py

@@ -0,0 +1,13 @@
+from ..globals import get_anyvar
+
+def get(vartype):
+    av = get_anyvar()
+    if vartype == "substitution":
+        out = av.object_store.substitution_count()
+    elif vartype == "deletion":
+        out = av.object_store.deletion_count()
+    elif vartype == "insertion":
+        out = av.object_store.insertion_count()
+    elif vartype == "all":
+        out = len(av.object_store)
+    return out, 200

src/anyvar/storage/postgres.py

@@ -54,7 +54,20 @@ def __del__(self):
         self._db.close()
 
     def __len__(self):
-        return self._db.__len__()
+        data = self.conn._fetchone(f"select count(*) as c from vrs_objects where vrs_object ->> 'type' = 'Allele'")
+        return data[0]
+
+    def deletion_count(self):
+        data = self.conn._fetchone(f"select count(*) as c from vrs_objects where length(vrs_object -> 'state' ->> 'sequence') = 0")
+        return data[0]
+
+    def substitution_count(self):
+        data = self.conn._fetchone(f"select count(*) as c from vrs_objects where length(vrs_object -> 'state' ->> 'sequence') = 1")
+        return data[0]
+
+    def insertion_count(self):
+        data = self.conn._fetchone(f"select count(*) as c from vrs_objects where length(vrs_object -> 'state' ->> 'sequence') > 1")
+        return data[0]
 
     def __iter__(self):
         return self._db.__iter__()