Merge pull request #672 from alliance-genome/mim_updates

OMIM to MIM updates.

README.md

@@ -1,4 +1,3 @@
-[![Build Status](https://travis-ci.org/alliance-genome/agr_loader.svg?branch=develop)](https://travis-ci.org/alliance-genome/agr_loader)
 [![Codacy Badge](https://api.codacy.com/project/badge/Grade/5259a0e847c04c72a4a9c4f34fabfed5)](https://www.codacy.com/project/christabone/agr_loader/dashboard?utm_source=github.com&utm_medium=referral&utm_content=alliance-genome/agr_loader&utm_campaign=Badge_Grade_Dashboard)
 
 #  Alliance of Genome Resources Loader
@@ -74,3 +73,4 @@ _Reminder_: authentification needs to be renewed every time you get an error lik
 ```
 Error response from daemon: pull access denied for 100225593120.dkr.ecr.us-east-1.amazonaws.com/agr_neo4j_env, repository does not exist or may require 'docker login': denied: Your authorization token has expired. Reauthenticate and try again.
 ```
+

src/etl/disease_etl.py

@@ -315,7 +315,7 @@ def process_pages(self, dp, xrefs, pages):
         for page in pages:
             if (self.data_provider == 'RGD' or self.data_provider == 'HUMAN') and prefix == 'DOID':
                 display_name = 'RGD'
-            elif (self.data_provider == 'RGD' or self.data_provider == 'HUMAN') and prefix == 'OMIM':
+            elif (self.data_provider == 'RGD' or self.data_provider == 'HUMAN') and (prefix == 'OMIM' or prefix == 'MIM'):
                 display_name = 'OMIM'
             else:
                 display_name = cross_ref_id.split(":")[0]

src/etl/helpers/etl_helper.py

@@ -154,7 +154,7 @@ def get_xref_dict(local_id, prefix, cross_ref_type, page,
     def get_complete_url_ont(self, local_id, global_id, key=None):
         """Get Complete 'ont'."""
         page = None
-        if 'OMIM:PS' in global_id:
+        if 'OMIM:PS' in global_id or 'MIM:PS' in global_id:
             page = 'ont'
 
         if not key:  # split not done before hand

src/etl/helpers/resource_descriptor_helper_2.py

@@ -302,7 +302,13 @@ def return_url(self, identifier, page):
     def return_url_from_identifier(self, identifier, page=None):
         """Return URL for an identifier."""
         db_prefix, identifier_stripped, separator = self.split_identifier(identifier)
-
+
+        # Normalize the identifier
+        # Special case for EFO.
+        if db_prefix and db_prefix.upper() == "EFO":
+            if not identifier_stripped.startswith("EFO_"):
+                identifier_stripped = "EFO_" + identifier_stripped
+
         key = self.get_key(db_prefix, identifier)
         if not key:
             return None

src/test/specific_tests.py

@@ -1168,7 +1168,7 @@ def test_human_dej_has_omim_full_url_cross_reference():
     """Test Human DEJ has OMIM Full URL Cross Reference"""
 
     query = """MATCH (g:Gene)--(dej:DiseaseEntityJoin)--(cr:CrossReference)
-               WHERE cr.crossRefCompleteUrl = 'https://www.omim.org/entry/605242'
+               WHERE cr.crossRefCompleteUrl = 'https://www.omim.org/MIM:605242'
                RETURN count(cr) AS counter"""
     with Neo4jHelper.run_single_query(query) as result:
         for record in result:

src/test/unit_tests.py

@@ -73,7 +73,7 @@ def test_url_lookup(self):
                     'result': 'https://ncit.nci.nih.gov/ncitbrowser/ConceptReport.jsp?dictionary=NCI_Thesaurus&code=C5604'},
                    {'local_id': 'badregexdoesnotmatch', 'global_id': 'MESH:badregexdoesnotmatch',
                     'result': 'https://www.ncbi.nlm.nih.gov/mesh/badregexdoesnotmatch'},
-                   {'local_id': 'Cdiff', 'global_id': 'MIM:1111', 'result': 'https://www.omim.org/entry/1111'}] 
+                   {'local_id': 'Cdiff', 'global_id': 'MIM:1111', 'result': 'https://www.omim.org/MIM:1111'}] 
 
         for item in lookups:
             url = self.etlh.get_complete_url_ont(item['local_id'], item['global_id'])