Merge branch 'master' into go-site-2002-gorule-0000061-update-for-new…

…-requirements

README-developers.md

@@ -46,12 +46,13 @@ git push --tags
 
 > To ensure this is successful, make sure you have relevant permissions to Ontobio package on [PyPI](https://pypi.org/project/ontobio/)
 
-Run the following commands:
+Run the following commands (note, you will need a pypi token to do this work, see: https://pypi.org/help/#apitoken
+and make a ~/.pypirc file to store your token):
 
 ```sh
 make cleandist
-python setup.py sdist bdist_wheel bdist_egg
-twine upload --repository-url https://upload.pypi.org/legacy/ --username PYPI_USERNAME dist/*
+python setup.py sdist bdist_wheel
+twine upload --repository-url https://upload.pypi.org/legacy/ --username __token__ dist/*
 ```
 
 ****************************************************************************************************

ontobio/__init__.py

@@ -1,6 +1,6 @@
 from __future__ import absolute_import
 
-__version__ = '2.8.11'
+__version__ = '2.8.14'
 
 
 from .ontol_factory import OntologyFactory

ontobio/io/gafparser.py

@@ -410,7 +410,10 @@ 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)

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"
 

ontobio/io/qc.py

@@ -646,6 +646,17 @@ def test(self, annotation: association.GoAssociation, config: assocparser.AssocP
 
         return result
 
+class GoRule51(GoRule):
+
+    def __init__(self):
+            super().__init__("GORULE:0000051", "Some GO terms require a value in the Annotation Extension field", FailMode.SOFT)
+
+    def test(self, annotation: association.GoAssociation, config: assocparser.AssocParserConfig, group=None) -> TestResult:
+
+        term = str(annotation.object.id)
+        no_extensions = (annotation.object_extensions is None or len(annotation.object_extensions) == 0)
+        fails = (term == "GO:0005488" or term == "GO:0005515") and no_extensions
+        return self._result(not fails)
 
 class GoRule55(GoRule):
 
@@ -919,6 +930,7 @@ def test(self, annotation: association.GoAssociation, config: assocparser.AssocP
     "GoRule43": GoRule43(),
     "GoRule46": GoRule46(),
     "GoRule50": GoRule50(),
+    "GoRule51": GoRule51(),   
     "GoRule55": GoRule55(),
     "GoRule57": GoRule57(),
     "GoRule58": GoRule58(),

tests/resources/errors.gaf

@@ -5,7 +5,7 @@
 !
 ! bad GO ID
 PomBase	SPAC25B8.17	ypf1		BAD_GO_ID	GO_REF:0000024	ISO	SGD:S000001583	C	intramembrane aspartyl protease of the perinuclear ER membrane Ypf1 (predicted)	ppp81	protein	taxon:4896	20150305	PomBase
-! No gene symbol
+! No gene symbol, fail since gene symbol is a required field
 PomBase	SPAC977.10			GO:1990578	PMID:19171118	IDA		C	plasma membrane sodium ion/proton antiporter Nhe1/Sod2	sod2	protein	taxon:4896	20151201	PomBase
 ! Blank lines
 

tests/test_gafparser.py

@@ -365,7 +365,7 @@ def test_errors_gaf():
             n_invalid_idspace += 1
     assert len(msgs) == 13
     assert n_invalid_idspace == 1
-    assert len(assocs) == 2
+    assert len(assocs) == 1
 
     w = GafWriter()
     w.write(assocs)

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 
+
+

tests/test_qc.py

@@ -563,6 +563,16 @@ def test_gorule50():
     assoc.evidence.type = Curie.from_str(iea_eco)
     test_result = qc.GoRule50().test(assoc, all_rules_config())
     assert test_result.result_type == qc.ResultType.PASS
+
+def test_gorule51():
+    assoc = make_annotation(goid="GO:0005515", extension="has_input(GO:0003674),occurs_in(CL:123456)").associations[0]
+
+    test_result = qc.GoRule51().test(assoc, all_rules_config())
+    assert test_result.result_type == qc.ResultType.PASS
+
+    assoc = make_annotation(goid="GO:0005488").associations[0]
+    test_result = qc.GoRule51().test(assoc, all_rules_config())
+    assert test_result.result_type == qc.ResultType.WARNING      
 
 def test_gorule55():
     a = ["blah"] * 15
@@ -727,7 +737,7 @@ def test_all_rules():
     assoc = gafparser.to_association(a).associations[0]
 
     test_results = qc.test_go_rules(assoc, config).all_results
-    assert len(test_results.keys()) == 24
+    assert len(test_results.keys()) == 25
     assert test_results[qc.GoRules.GoRule26.value].result_type == qc.ResultType.PASS
     assert test_results[qc.GoRules.GoRule29.value].result_type == qc.ResultType.PASS