Merge pull request #645 from biolink/go-site-2066-gorule-0000001-inva…

…lid-taxon-id-0 Go site 2066 gorule 0000001 invalid taxon id 0
biolink · Sep 13, 2023 · 7d53dd3 · 7d53dd3
2 parents 5d15999 + d721f9d
commit 7d53dd3
Show file tree

Hide file tree

Showing 3 changed files with 94 additions and 1 deletion.
diff --git a/ontobio/io/gafparser.py b/ontobio/io/gafparser.py
@@ -401,7 +401,10 @@ def to_association(gaf_line: List[str], report=None, group="unknown", dataset="u
         return assocparser.ParseResult(source_line, [], True, report=report)
 
     taxon = parsed_taxons_result.parsed[0]
-
+    if taxon.identity is None or taxon.identity == '0':
+        report.error(source_line, Report.INVALID_TAXON, parsed_taxons_result.original, parsed_taxons_result.message, taxon=parsed_taxons_result.original, rule=1)
+        return assocparser.ParseResult(source_line, [], True, report=report)
+
     date = assocparser.parse_date(gaf_line[13], report, source_line)
     if date is None:
         return assocparser.ParseResult(source_line, [], True, report=report)

diff --git a/ontobio/io/gpadparser.py b/ontobio/io/gpadparser.py
@@ -294,6 +294,12 @@ def from_1_2(gpad_line: List[str], report=None, group="unknown", dataset="unknow
     if entity is not None:
         subject = entity
         taxon = subject.taxon
+
+    #Ensure taxon is valid, if we are reading from bioentity
+    if len(bio_entities.entities) > 0:
+        if taxon.identity is None or taxon.identity == '0':
+            report.error(source_line, Report.INVALID_TAXON, "None or 0", "Taxon is invalid", rule=1)
+            return assocparser.ParseResult(source_line, [], True, report=report)    
 
     go_term = association.Curie.from_str(gpad_line[3])
     if go_term.is_error():
@@ -425,6 +431,12 @@ def from_2_0(gpad_line: List[str], report=None, group="unknown", dataset="unknow
         # If we found a subject entity, then set `subject` to the found entity
         subject = entity
         taxon = subject.taxon
+
+    #Ensure taxon is valid, if we are reading from bioentity
+    if len(bio_entities.entities) > 0:
+        if taxon.identity is None or taxon.identity == '0':
+            report.error(source_line, Report.INVALID_TAXON, "None or 0", "Taxon is invalid", rule=1)
+            return assocparser.ParseResult(source_line, [], True, report=report)     
 
     negated = gpad_line[1] == "NOT"
 

diff --git a/tests/test_gpad_parser.py b/tests/test_gpad_parser.py
@@ -3,6 +3,8 @@
 from ontobio.model import association
 from ontobio.model.association import ConjunctiveSet, ExtensionUnit, Curie
 from ontobio.ontol_factory import OntologyFactory
+from ontobio.model import collections
+from ontobio.model.association import Curie, Subject
 
 import yaml
 
@@ -252,3 +254,79 @@ def test_unmapped_eco_to_gaf_codes():
     vals[5] = "ECO:0006003"  # indirectly maps to IDA via gaf-eco-mapping-derived.txt
     result = parser.parse_line("\t".join(vals))
     assert len(result.associations) == 1
+
+def test_gpi_check():
+    report = assocparser.Report(group="unknown", dataset="unknown")
+    vals = [
+        "ZFIN",
+        "ZDB-GENE-070117-1552",
+        "acts_upstream_of_or_within",
+        "GO:0045601",
+        "PMID:17531218",
+        "ECO:0000307",
+        "",
+        "",
+        "20080326",
+        "ZFIN",
+        "",
+        "creation-date=2020-09-17|modification-date=2020-09-17|contributor-id=http://orcid.org/0000-0003-2689-5511"
+    ]
+
+    bioentities = {'ZFIN:ZDB-GENE-070117-1552'  : {
+            'id': "ZDB-GENE-070117-1552",
+            'label': "ZDB-GENE-070117-1552",
+            'full_name': "fullnames",
+            'synonyms': "synonyms",
+            'type': "gene_product",
+            'taxon': "taxon:0"
+        }
+        }
+
+    bioentities = collections.BioEntities({
+        Curie("ZFIN", "ZDB-GENE-070117-1552"): Subject(Curie.from_str("ZFIN:ZDB-GENE-070117-1552"), "ste4", ["adaptor protein Ste4"], [], ["protein"], Curie.from_str("taxon:0"))
+        })
+
+
+
+    result = to_association(list(vals), report=report, version="1.2", bio_entities=bioentities)
+    assert result.skipped == True
+    assert len([m for m in result.report.messages if m["level"] == "ERROR"]) == 1
+    assert len(result.associations) == 0 
+
+    vals = [
+        "ZFIN:ZDB-GENE-070117-1552",
+        "ZFIN:ZDB-GENE-070117-1552",
+        "RO:12345",
+        "GO:0045601",
+        "PMID:17531218",
+        "ECO:0000307",
+        "",
+        "",
+        "2008-03-26",
+        "ZFIN",
+        "",
+        "creation-date=2020-09-17|modification-date=2020-09-17|contributor-id=http://orcid.org/0000-0003-2689-5511"
+    ]
+    result = to_association(list(vals), report=report, version="2.0", bio_entities=bioentities)
+    assert result.skipped == True
+    assert len([m for m in result.report.messages if m["level"] == "ERROR"]) == 2
+    assert len(result.associations) == 0     
+
+
+    bioentities = collections.BioEntities({"bla": 'blabla'})
+
+    result = to_association(list(vals), report=report, version="2.0", bio_entities=bioentities)
+    assert result.skipped == True
+    assert len([m for m in result.report.messages if m["level"] == "ERROR"]) == 3
+    assert len(result.associations) == 0    
+
+    bioentities = collections.BioEntities({
+        Curie("ZFIN", "ZDB-GENE-070117-1552"): Subject(Curie.from_str("ZFIN:ZDB-GENE-070117-1552"), "ste4", ["adaptor protein Ste4"], [], ["protein"], Curie.from_str("NCBITaxon:12345"))
+        })
+
+    result = to_association(list(vals), report=report, version="2.0", bio_entities=bioentities)
+    assert result.skipped == 0
+    assert len([m for m in result.report.messages if m["level"] == "ERROR"]) == 3
+    assert len(result.associations) == 1 
+
+