version 0.1.0 finalizada

Author: mauri101-Ar

.gitignore

@@ -10,4 +10,5 @@ docker-compose.yml
 bio-api/secretkey.txt
 *.log
 databases/hgnc/hgnc_output.json
-databases/hgnc/hgnc_dataset.tsv
+databases/hgnc/hgnc_dataset.tsv
+databases/cpdb/

bio-api/bioapi.py

@@ -6,11 +6,12 @@
 from flask import request
 import urllib.parse
 import pymongo
+from pymongo.database import Database
 import re
 import os
 import configparser
 import logging
-from pymongo.database import Database
+
 
 # Gets production flag
 IS_DEBUG: bool = os.environ.get('DEBUG', 'true') == 'true'
@@ -60,7 +61,7 @@ def get_mongo_connection() -> Database:
         mongo_client = pymongo.MongoClient(f"mongodb://{user}:{password}@{host}:{port}/?authSource=admin")
         return mongo_client[db]
     except Exception as e:
-        logging.error("BioAPI no se pudo conectar a la base de datos MongoDB configurada." + str(e), exc_info=True)
+        logging.error("Database connection error." + str(e), exc_info=True)
         exit(-1)
 
 
@@ -151,8 +152,14 @@ def genSymbol(gene_id):
         respuesta = {gene_id: []}
         try:
             gv = mapear_gen(gene_id)
+            if len(gv) == 0:
+                abort(404, "invalid gene identifier")
             respuesta[gene_id] = gv
-        except Exception as e:
+        except TypeError as e:
+            abort(400, e)
+        except ValueError as e:
+            abort(400, e)
+        except KeyError as e:
             abort(400, e)
         return make_response(respuesta, 200, headers)
 
@@ -178,10 +185,9 @@ def genSymbols():
     def genes_of_the_same_family(gene_id):
         respuesta = {"gene_id": None, "groups": [], "locus_group": None, "locus_type": None}
         try:
-            
             mapped_gene = mapear_gen(gene_id)
             if len(mapped_gene) == 0:
-                abort(400, "invalid gene identifier")
+                abort(404, "invalid gene identifier")
             elif len(mapped_gene) >= 2:
                 abort(400, "ambiguous gene identifier. The identifier may refer to more than one HGNC-approved gene (" + ",".join(mapped_gene) + ")")
             approved_symbol = mapped_gene[0]
@@ -208,25 +214,13 @@ def genes_of_the_same_family(gene_id):
 
     # Manejo de errores
     @flask_app.errorhandler(400)
-    def resource_not_found(e):
+    def bad_request(e):
         return jsonify(error=str(e)), 400
 
-    @flask_app.errorhandler(405)
-    def resource_not_found(e):
-        return jsonify(error=str(e)), 405
-
     @flask_app.errorhandler(404)
-    def resource_not_found(e):
+    def not_found(e):
         return jsonify(error=str(e)), 404
 
-    @flask_app.errorhandler(409)
-    def resource_not_found(e):
-        return jsonify(error=str(e)), 409
-
-    @flask_app.errorhandler(500)
-    def resource_not_found(e):
-        return jsonify(error=str(e)), 500
-
     return flask_app
 
 

bio-api/tests/test_gene_symbol.py

@@ -8,7 +8,6 @@ def test_gene_id_valid(client):
     response = client.get(f'{URL_BASE}/{valid_id}')
     res = json.loads(response.data)
     assert response.status_code == 200
-    assert valid_id in list(res.keys())
     assert type(res[valid_id]) == list
     assert "TP53" in res[valid_id]
 
@@ -18,7 +17,6 @@ def test_gene_alias_two_genes(client):
     response = client.get(f'{URL_BASE}/{alias_id}')
     res = json.loads(response.data)
     assert response.status_code == 200
-    assert alias_id in list(res.keys())
     assert type(res[alias_id]) == list
     assert "BRCA1" in res[alias_id]
     assert "ICE2" in res[alias_id]
@@ -29,15 +27,12 @@ def test_gene_id_invalid(client):
     invalid_id="XXX" #Invalid id
     response = client.get(f'{URL_BASE}/{invalid_id}')
     res = json.loads(response.data)
-    assert response.status_code == 200
-    assert invalid_id in list(res.keys())
-    assert type(res[invalid_id]) == list
-    assert res[invalid_id] == []
+    assert response.status_code == 404
+    assert "error" in list(res.keys())
 
 def test_missing_gene_id_param(client):
     """Tests missing mandatory parameter"""
     response = client.get(f'{URL_BASE}/')
     res = json.loads(response.data)
     assert response.status_code == 404
     assert "error" in list(res.keys())
-    assert type(res["error"]) == str

bio-api/tests/test_genes_same_group.py

@@ -8,21 +8,14 @@ def test_gene_in_group(client):
     response = client.get(f'{URL_BASE}/{valid_id}')
     res = json.loads(response.data)
     assert response.status_code == 200
-    assert "gene_id" in list(res.keys())
-    assert "groups" in list(res.keys())
-    assert "locus_group" in list(res.keys())
-    assert "locus_type" in list(res.keys())
-    assert type(res["gene_id"]) == str
     assert type(res["groups"]) == list
-    assert type(res["locus_group"]) == str
-    assert type(res["locus_type"]) == str
-    assert res["gene_id"] == "EGFR"
-    assert res["locus_group"] == "protein-coding gene"
-    assert res["locus_type"] == "gene with protein product"
     assert len(res["groups"]) == 1
     assert res["groups"][0]["gene_group"] == "Erb-b2 receptor tyrosine kinases"
     assert res["groups"][0]["gene_group_id"] == "1096"
     assert sorted(res["groups"][0]["genes"]) == sorted(["ERBB3","EGFR","ERBB2","ERBB4"])
+    assert res["gene_id"] == "EGFR"
+    assert res["locus_group"] == "protein-coding gene"
+    assert res["locus_type"] == "gene with protein product"
 
 
 def test_gene_not_in_group(client):
@@ -31,19 +24,12 @@ def test_gene_not_in_group(client):
     response = client.get(f'{URL_BASE}/{valid_id}')
     res = json.loads(response.data)
     assert response.status_code == 200
-    assert "gene_id" in list(res.keys())
-    assert "groups" in list(res.keys())
-    assert "locus_group" in list(res.keys())
-    assert "locus_type" in list(res.keys())
-    assert type(res["gene_id"]) == str
     assert type(res["groups"]) == list
-    assert type(res["locus_group"]) == str
-    assert type(res["locus_type"]) == str
     assert res["gene_id"] == "TP53"
     assert res["locus_group"] == "protein-coding gene"
     assert res["locus_type"] == "gene with protein product"
     assert len(res["groups"]) == 0
-
+ 
 
 def test_gene_alias(client):
     """test with gene symbol that is alias of two valid symbols"""
@@ -52,7 +38,6 @@ def test_gene_alias(client):
     res = json.loads(response.data)
     assert response.status_code == 400
     assert "error" in list(res.keys())
-    assert type(res["error"]) == str
     assert res["error"] == "400 Bad Request: ambiguous gene identifier. The identifier may refer to more than one HGNC-approved gene (BRCA1,ICE2)"
 
 
@@ -62,8 +47,6 @@ def test_invalid_gene_symbol(client):
     response = client.get(f'{URL_BASE}/{invalid_id}')
     res = json.loads(response.data)
     assert response.status_code == 400
-    assert "error" in list(res.keys())
-    assert type(res["error"]) == str
     assert res["error"] =="400 Bad Request: invalid gene identifier"
 
 
@@ -72,6 +55,4 @@ def test_invalid_gene_symbol(client):
     response = client.get(f'{URL_BASE}')
     res = json.loads(response.data)
     assert response.status_code == 404
-    assert "error" in list(res.keys())
-    assert type(res["error"]) == str
     assert res["error"] == "404 Not Found: The requested URL was not found on the server. If you entered the URL manually please check your spelling and try again."

bio-api/tests/test_genes_symbols.py

@@ -13,25 +13,20 @@ def test_gene_id_valid(client):
     }
     response = client.post(f'{URL_BASE}', data=json.dumps(data), headers=headers)
     res = json.loads(response.data)
-    assert response.content_type == 'application/json'
     assert response.status_code == 200
-    assert valid_id in list(res.keys())
     assert type(res[valid_id]) == list
     assert "BRAF" in res[valid_id]
 
 def test_valid_and_invalid_genes(client):
-    """"test for one valid and one invalid gene"""
+    """test for one valid and one invalid gene"""
     valid_id="HGNC:3236"    #hgnc_id
     invalid_id="blcdtm"     #invalid id
     data = {    
         "genes_ids" : [ valid_id, invalid_id ]
     }
     response = client.post(f'{URL_BASE}', data=json.dumps(data), headers=headers)
     res = json.loads(response.data)
-    assert response.content_type == 'application/json'
     assert response.status_code == 200
-    assert valid_id in list(res.keys())
-    assert invalid_id in list(res.keys())
     assert type(res[valid_id]) == list
     assert type(res[invalid_id]) == list
     assert res[valid_id][0] == "EGFR"
@@ -47,10 +42,7 @@ def test_gene_alias_two_genes_and_invalid_id(client):
     }
     response = client.post(f'{URL_BASE}', data=json.dumps(data), headers=headers)
     res = json.loads(response.data)
-    assert response.content_type == 'application/json'
     assert response.status_code == 200
-    assert alias_id in list(res.keys())
-    assert invalid_id in list(res.keys())
     assert type(res[alias_id]) == list
     assert type(res[invalid_id]) == list
     assert res[invalid_id] == []
@@ -68,7 +60,6 @@ def test_empty_gene_id(client):
     }
     response = client.post(f'{URL_BASE}', data=json.dumps(data), headers=headers)
     res = json.loads(response.data)
-    assert response.content_type == 'application/json'
     assert response.status_code == 200
     assert len(list(res.keys())) == 0
     assert res == {}
@@ -81,7 +72,5 @@ def test_corrupted_structure(client):
     }
     response = client.post(f'{URL_BASE}', data=json.dumps(data), headers=headers)
     res = json.loads(response.data)
-    assert response.content_type == 'application/json'
     assert response.status_code == 400
-    assert "error" in list(res.keys())
-    assert res["error"] == "400 Bad Request: genes_ids is mandatory"
+    assert "error" in list(res.keys())