Merge pull request #702 from biolink/pipeline-408-gorule-0000027-mgi-…

…failing-go-checks For geneontology/pipeline#408
biolink · Dec 14, 2024 · 5a3da4f · 5a3da4f
2 parents 85dce88 + 69e0b37
commit 5a3da4f
Show file tree

Hide file tree

Showing 3 changed files with 82 additions and 4 deletions.
diff --git a/ontobio/io/assocparser.py b/ontobio/io/assocparser.py
@@ -800,9 +800,11 @@ def _validate_id(self, id, line: SplitLine, allowed_ids=None, context=None):
             return False
 
         (id_prefix, right) = id.split(":", maxsplit=1)
+        mgi_id = None
         if right.startswith("MGI:"):
             ## See ticket https://github.com/geneontology/go-site/issues/91
             ## For purposes of determining allowed IDs in DB XREF, MGI IDs shall look like `MGI:12345`
+            mgi_id = right
             right = right[4:]
 
         if id_prefix == "" or right == "":
@@ -830,9 +832,16 @@ def _validate_id(self, id, line: SplitLine, allowed_ids=None, context=None):
                         if regex.fullmatch(right):
                             identity_matches_pattern = True
                             break
-                    if identity_matches_pattern == False:
+                        # check syntax for mgi using id instead of internal representation
+                        if mgi_id is not None and regex.fullmatch(mgi_id):
+                            identity_matches_pattern = True
+                            break
+                    if identity_matches_pattern == False and mgi_id is None:
                         self.report.warning(line.line, Report.INVALID_ID, id,
                         "GORULE:0000027: {} does not match any id_syntax patterns for {} in dbxrefs".format(right, id_prefix), taxon=line.taxon, rule=27)
+                    elif identity_matches_pattern == False and mgi_id is not None:
+                        self.report.warning(line.line, Report.INVALID_ID, id,
+                        "GORULE:0000027: {} does not match any id_syntax patterns for {} in dbxrefs".format(mgi_id, id_prefix), taxon=line.taxon, rule=27)    
             else:
                 self.report.warning(line.line, Report.INVALID_ID, id,
                     "GORULE:0000027: {} not found in list of database names in dbxrefs".format(id_prefix), taxon=line.taxon, rule=27)  

diff --git a/tests/test_gafparser.py b/tests/test_gafparser.py
@@ -633,7 +633,11 @@ def test_id_syntax():
     database_id_syntax_lookups['PomBase'] = pombase_types
 
     wb_ref_types = {}
-    database_id_syntax_lookups['WB_REF'] = wb_ref_types 
+    database_id_syntax_lookups['WB_REF'] = wb_ref_types
+
+    mgi_types = {}
+    mgi_types['entity'] = re.compile('MGI:[0-9]{5,}')
+    database_id_syntax_lookups['MGI'] = mgi_types
 
     p = GafParser(config=assocparser.AssocParserConfig(
         ontology=OntologyFactory().create(ONT), db_type_name_regex_id_syntax=database_id_syntax_lookups))
@@ -648,7 +652,13 @@ def test_id_syntax():
     assert len(assoc_result.associations) == 1
     assert assoc_result.skipped == False
     messages = p.report.to_report_json()["messages"]
-    assert "gorule-0000027" not in messages    
+    assert "gorule-0000027" not in messages
+
+    assoc_result = p.parse_line("PomBase\tSPBC1289.03c\tspi1\t\tGO:0005515\tWB_REF:WBPaper00006408|PMID:18422602\tIPI\tMGI:MGI:1298204\tF\tRan GTPase Spi1\t\tprotein\ttaxon:4896\t20080718\tPomBase\t")
+    assert len(assoc_result.associations) == 1
+    assert assoc_result.skipped == False
+    messages = p.report.to_report_json()["messages"]
+    assert "gorule-0000027" not in messages        
 
     p = GafParser(config=assocparser.AssocParserConfig(
         ontology=OntologyFactory().create(ONT), db_type_name_regex_id_syntax=database_id_syntax_lookups))
@@ -686,6 +696,15 @@ def test_id_syntax():
     assert len(messages["gorule-0000027"]) == 1
     assert messages["gorule-0000027"][0]["obj"] == "BLA:18422602" 
 
+    p = GafParser(config=assocparser.AssocParserConfig(
+        ontology=OntologyFactory().create(ONT), db_type_name_regex_id_syntax=database_id_syntax_lookups))    
+    assoc_result = p.parse_line("PomBase\tSPBC1289.03c\tspi1\t\tGO:0005515\tWB_REF:WBPaper00006408|PMID:18422602\tIPI\tMGI:1298204\tF\tRan GTPase Spi1\t\tprotein\ttaxon:4896\t20080718\tPomBase\t")
+    assert len(assoc_result.associations) == 1
+    assert assoc_result.skipped == False
+    messages = p.report.to_report_json()["messages"]
+    assert len(messages["gorule-0000027"]) == 1
+    assert messages["gorule-0000027"][0]["obj"] == "MGI:1298204"  
+
 
 def test_gaf_gpi_bridge():
     gaf = ["MGI", "MGI:1923503", "0610006L08Rik", "enables", "GO:0003674", "MGI:MGI:2156816|GO_REF:0000015", "ND", "",

diff --git a/tests/test_gpad_parser.py b/tests/test_gpad_parser.py
@@ -345,6 +345,10 @@ def test_id_syntax():
     pombase_types['entity'] = re.compile('S\\w+(\\.)?\\w+(\\.)?')
     database_id_syntax_lookups['PomBase'] = pombase_types
 
+    mgi_types = {}
+    mgi_types['entity'] = re.compile('MGI:[0-9]{5,}')
+    database_id_syntax_lookups['MGI'] = mgi_types    
+
     eco_types = {}
     eco_types['entity'] = re.compile(pattern)
     database_id_syntax_lookups['ECO'] = eco_types
@@ -396,7 +400,31 @@ def test_id_syntax():
     assert len(result.associations) == 1
     assert result.skipped == False  
     messages = p.report.to_report_json()["messages"]
-    assert "gorule-0000027" not in messages     
+    assert "gorule-0000027" not in messages
+
+    vals = ["PomBase",
+            "SPAC25A8.01c",
+            "acts_upstream_of_or_within",
+            "GO:0007155",
+            "MGI:MGI:1298204",
+            "ECO:0000305",
+            "GO:0005913",
+            "",
+            "20041026",
+            "ZFIN",
+            "",
+            "PomBase"
+            ]
+
+    config = assocparser.AssocParserConfig(
+        ontology=OntologyFactory().create(ALT_ID_ONT), db_type_name_regex_id_syntax=database_id_syntax_lookups)
+    p = GpadParser(config=config)
+    result = p.parse_line("\t".join(vals))
+    assert len(result.associations) == 1
+    assert result.skipped == False  
+    messages = p.report.to_report_json()["messages"]
+    assert "gorule-0000027" not in messages    
+
 
     vals = ["PomBase",
             "SPAC25A8.01c",
@@ -487,6 +515,28 @@ def test_id_syntax():
     assert len(messages["gorule-0000027"]) == 1
     assert messages["gorule-0000027"][0]["obj"] == "BLA:15494018"   
 
+    vals = ["PomBase",
+            "SPAC25A8.01c",
+            "acts_upstream_of_or_within",
+            "GO:0007155",
+            "MGI:15494018",
+            "ECO:0000305",
+            "GO:0005913",
+            "",
+            "20041026",
+            "ZFIN",
+            "",
+            "PomBase"
+            ]
+    p = GpadParser(config=config)
+    result = p.parse_line("\t".join(vals))
+    assert len(result.associations) == 1
+    assert result.skipped == False
+    messages = p.report.to_report_json()["messages"]
+    assert len(messages["gorule-0000027"]) == 1
+    assert messages["gorule-0000027"][0]["obj"] == "MGI:15494018"       
+
+
 def test_gpi_check():
     report = assocparser.Report(group="unknown", dataset="unknown")
     vals = [