def __init__(self, config=None, group="unknown", dataset="unknown"):
"""
Arguments:
---------
config : a AssocParserConfig object
"""
if config is None:
config = assocparser.AssocParserConfig()
self.config = config
self.report = assocparser.Report(group=group, dataset=dataset)
self.gpi = None
if self.config.gpi_authority_path is not None:
self.gpi = dict()
parser = entityparser.GpiParser()
with open(self.config.gpi_authority_path) as gpi_f:
entities = parser.parse(file=gpi_f)
for entity in entities:
self.gpi[entity["id"]] = {
"symbol": entity["label"],
"name": entity["full_name"],
"synonyms": entitywriter.stringify(entity["synonyms"]),
"type": entity["type"]
}
print("Loaded {} entities from {}".format(len(self.gpi.keys()), self.config.gpi_authority_path))
def __init__(self, config=None, group="unknown", dataset="unknown", bio_entities=None):
"""
Arguments:
---------
config : a AssocParserConfig object
"""
self.config = config
self.group = group
self.version = None
self.default_version = "2.1"
self.bio_entities = bio_entities
if self.bio_entities is None:
self.bio_entities = collections.BioEntities(dict())
self.cell_component_descendants_closure = None
if config is None:
self.config = assocparser.AssocParserConfig()
self.report = assocparser.Report(group=group, dataset=dataset, config=self.config)
# self.gpi = None
if self.config.gpi_authority_path is not None:
gpi_paths = self.config.gpi_authority_path
if isinstance(gpi_paths, str):
gpi_paths = [gpi_paths]
for gpi_path in gpi_paths:
gpi_bio_entities = collections.BioEntities.load_from_file(gpi_path)
self.bio_entities.merge(gpi_bio_entities)
print("Loaded {} entities from {}".format(len(gpi_bio_entities.entities.keys()), gpi_path))
def test_model_dates():
model_associations = []
version = "2.0"
report = assocparser.Report(group="unknown", dataset="unknown")
vals = [
"MGI:MGI:1915834", "", "RO:0002327", "GO:0003674",
"MGI:MGI:2156816|GO_REF:0000015", "ECO:0000307", "", "", "2020-10-09",
"MGI", "",
"creation-date=2020-09-17|modification-date=2020-10-09|contributor-id=http://orcid.org/0000-0003-2689-5511"
]
model_associations = model_associations + to_association(
list(vals), report=report, version=version).associations
# Different term, earlier creation-date
vals[3], vals[11] = "GO:0016301", "creation-date=2011-12-13"
model_associations = model_associations + to_association(
list(vals), report=report, version=version).associations
# Different term, no annotation properties
vals[3], vals[11] = "GO:0001962", ""
model_associations = model_associations + to_association(
list(vals), report=report, version=version).associations
builder = gocam_builder.GoCamBuilder(parser_config=PARSER_CONFIG,
modelstate="test")
model = builder.translate_to_model(gene="MGI:MGI:1915834",
assocs=model_associations)
assert model.date == "2020-10-09"
assert model.creation_date == "2011-12-13"
assert model.import_date == datetime.date.today().isoformat()
def __init__(self,
config=assocparser.AssocParserConfig(),
group="unknown",
dataset="unknown",
bio_entities=None):
"""
Arguments:
---------
config : a AssocParserConfig object
"""
self.config = config
self.report = assocparser.Report(config=self.config,
group="unknown",
dataset="unknown")
self.version = None
self.default_version = "1.2"
self.bio_entities = bio_entities
if self.bio_entities is None:
self.bio_entities = collections.BioEntities(dict())
if self.config.gpi_authority_path is not None:
self.bio_entities.merge(
collections.BioEntities.load_from_file(
self.config.gpi_authority_path))
print("Loaded {} entities from {}".format(
len(self.bio_entities.entities.keys()),
self.config.gpi_authority_path))
def __init__(self, config=assocparser.AssocParserConfig()):
"""
Arguments:
---------
config : a AssocParserConfig object
"""
self.config = config
self.report = assocparser.Report(config=self.config)
def test_parse_2_0():
version = "2.0"
report = assocparser.Report(group="unknown", dataset="unknown")
vals = [
"MGI:MGI:1918911",
"",
"RO:0002327",
"GO:0003674",
"MGI:MGI:2156816|GO_REF:0000015",
"ECO:0000307",
"",
"",
"2020-09-17",
"MGI",
"",
"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=version)
assert result.skipped == 0
assert len([m for m in result.report.messages if m["level"] == "ERROR"]) == 0
assert len(result.associations) == 1
# Annotation_Extensions
vals[10] = "BFO:0000066(CL:0000010),GOREL:0001004(CL:0000010)"
result = to_association(list(vals), report=report, version=version)
assert result.associations[0].object_extensions == [ConjunctiveSet([
ExtensionUnit(Curie("BFO", "0000066"), Curie("CL", "0000010")),
ExtensionUnit(Curie("GOREL", "0001004"), Curie("CL", "0000010"))
])]
# With_or_From
vals[6] = "PR:Q505B8|PR:Q8CHK4"
result = to_association(list(vals), report=report, version=version)
assert result.associations[0].evidence.with_support_from == [
ConjunctiveSet([Curie("PR", "Q505B8")]),
ConjunctiveSet([Curie("PR", "Q8CHK4")])
]
# Interacting taxon - this example should fail
vals[7] = "Staphylococcus aureus ; NCBITaxon:1280"
result = to_association(list(vals), report=report, version=version)
assert len([m for m in result.report.messages if m["level"] == "ERROR"]) > 0
assert len(result.associations) == 0
# Now test valid interacting taxon value
vals[7] = "NCBITaxon:1280"
result = to_association(list(vals), report=report, version=version)
assert result.associations[0].interacting_taxon == Curie("NCBITaxon", "1280")
# Confirm non-"MGI:MGI:" IDs will parse
vals[0] = "WB:WBGene00001189"
result = to_association(list(vals), report=report, version=version)
assert result.associations[0].subject.id == Curie("WB", "WBGene00001189")
# Test annotation property retrieval
contributors = result.associations[0].annotation_property_values(property_key="contributor-id")
assert set(contributors) == {"http://orcid.org/0000-0003-2689-5511"}
def __init__(self, config=None):
"""
Arguments:
---------
config : a AssocParserConfig object
"""
if config == None:
config = assocparser.AssocParserConfig()
self.config = config
self.report = assocparser.Report()
def test_parse_interacting_taxon():
report = assocparser.Report(group="unknown", dataset="unknown")
vals = [
"MGI", "MGI:1918911", "enables", "GO:0003674",
"MGI:MGI:2156816|GO_REF:0000015", "ECO:0000307", "", "taxon:5678",
"20100209", "MGI", "",
"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="1.2")
assert result.associations[0].interacting_taxon == Curie(
namespace="NCBITaxon", identity="5678")
def __init__(self, config=None):
"""
Arguments:
---------
config : a assocparser.AssocParserConfig object
"""
if config is None:
config = assocparser.AssocParserConfig()
self.config = config
self.report = assocparser.Report()
self.default_version = "1.2"
self.version = None
def test_parse_1_2():
report = assocparser.Report(group="unknown", dataset="unknown")
vals = [
"MGI", "MGI:1918911", "enables", "GO:0003674",
"MGI:MGI:2156816|GO_REF:0000015", "ECO:0000307", "", "", "20100209",
"MGI", "",
"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="1.2")
assert result.skipped == 0
assert len([m for m in result.report.messages
if m["level"] == "ERROR"]) == 0
assert len(result.associations) == 1
def test_get_with_froms():
gpi_ents = gocam_builder.GoCamBuilder.parse_gpi(
gpi_file="tests/resources/mgi2.test_entities.gpi")
report = assocparser.Report(group="unknown", dataset="unknown")
vals = [
"MGI:MGI:1929608",
"",
"RO:0001025",
"GO:0005515",
"MGI:MGI:3028747|PMID:14627703",
"ECO:0000353",
"MGI:MGI:1915834", # with/from
"",
"2004-04-30",
"MGI",
"",
"creation-date=2004-04-30|modification-date=2004-04-30|contributor-id=http://orcid.org/0000-0002-9796-7693"
]
result = to_association(list(vals), report=report, version="2.0")
go_assoc = result.associations[0]
ca_set = collapsed_assoc.CollapsedAssociationSet(GO_ONTO, gpi_ents)
with_froms = ca_set.get_with_froms(go_assoc)
assert len(with_froms) == 1 and \
with_froms[0].header == ["MGI:MGI:1915834"] and \
with_froms[0].line == ["MGI:MGI:1915834"]
vals[6] = "MGI:MGI:1915834|FAKE:12345"
result = to_association(list(vals), report=report, version="2.0")
go_assoc = result.associations[0]
ca_set = collapsed_assoc.CollapsedAssociationSet(GO_ONTO, gpi_ents)
with_froms = ca_set.get_with_froms(go_assoc)
# FAKE:12345 should be on line since not in GPI nor does it have same taxon as subject MGI:MGI:1929608
assert len(with_froms) == 2 and \
with_froms[0].header == ["MGI:MGI:1915834"] and \
with_froms[0].line == ["MGI:MGI:1915834"] and \
with_froms[1].header == [] and \
with_froms[1].line == ["FAKE:12345"]
# Test merging of same-header with/from values in different order
ca_set = collapsed_assoc.CollapsedAssociationSet(GO_ONTO, gpi_ents)
header1 = collapsed_assoc.GoAssocWithFrom(
header=["MGI:MGI:1915834", "FAKE:12345"])
header2 = collapsed_assoc.GoAssocWithFrom(
header=["FAKE:12345", "MGI:MGI:1915834"])
ca_set.find_or_create_collapsed_association(go_assoc, with_from=header1)
ca_set.find_or_create_collapsed_association(go_assoc, with_from=header2)
assert len(ca_set.collapsed_associations
) == 1 and ca_set.collapsed_associations[0].with_froms == [
"FAKE:12345", "MGI:MGI:1915834"
]
def extract_properties(annot: GoAssociation):
annotation_properties = {}
property_keys = set([prop[0] for prop in annot.properties])
for pk in property_keys:
property_values = annot.annotation_property_values(pk)
# Check for date fields (ex: "creation-date", "modification-date") and parse to Date
if pk.endswith("date"):
# Stealing some assocparser stuff here
dummy_report = assocparser.Report()
annot_line = annot.source_line.split("\t")
property_values = [
assocparser.parse_iso_date(v, dummy_report, annot_line)
for v in property_values
]
annotation_properties[pk] = property_values
return annotation_properties
def __init__(self,
config=assocparser.AssocParserConfig(),
group="unknown",
dataset="unknown",
bio_entities=None):
"""
Arguments:
---------
config : a AssocParserConfig object
"""
self.config = config
self.report = assocparser.Report(config=self.config,
group="unknown",
dataset="unknown")
self.version = None
self.default_version = "1.2"
self.bio_entities = bio_entities
if self.bio_entities is None:
self.bio_entities = collections.BioEntities(dict())
class GeneralAssocParser(assocparser.AssocParser):
config: assocparser.AssocParserConfig
group: str
headers: List[str] = field(default_factory=list)
report: assocparser.Report = field(
default_factory=lambda: assocparser.Report())
bio_entities: BioEntities = field(
default_factory=lambda: BioEntities(dict()))
annotation_parser: Optional[assocparser.AssocParser] = None
def parse_line(self, line):
parsed = super().validate_line(line)
if parsed:
return parsed
if self.is_header(line):
if self.annotation_parser is None:
# We're still looking for a version identifier
parser = create_parser_from_header(
line, self.config, bio_entities=self.bio_entities)
if parser is not None:
self.annotation_parser = parser
self.report = parser.report
self.headers.append(line)
return assocparser.ParseResult(line, [], skipped=False)
# At this point, we are not going through headers, and so we should have selected a parser
if self.annotation_parser is None:
logger.error("File is bad! We need to bail here!")
self.report.error("",
"Invalid Annotation File",
"",
"File header is missing file and version",
rule=1)
raise NoVersionInFile("File has no version info", self.report)
# Just hand off parse responsibility to underlying `annotation_parser`
return self.annotation_parser.parse_line(line)
def construct_collection(
annotation_path: Optional[str], gpi_paths: List[str],
config: assocparser.AssocParserConfig) -> AssociationCollection:
entities = BioEntities(dict())
for gpi in gpi_paths:
entities.merge(BioEntities.load_from_file(gpi))
annotations = []
headers = []
report = assocparser.Report()
if annotation_path:
parser = GeneralAssocParser(config, "unknown", bio_entities=entities)
try:
annotations = parser.parse(annotation_path)
except NoVersionInFile as e:
parser.report = e.report
annotations = []
headers = parser.headers
report = parser.report
return AssociationCollection(headers, GoAssociations(annotations),
entities, report)
def initial(): # type AssociationCollection
return AssociationCollection([], GoAssociations([]),
BioEntities(dict()), assocparser.Report())
