def produce_ttl(dataset, target_dir, gaf_path, ontology_graph):
gafparser = GafParser()
gafparser.config = assocparser.AssocParserConfig(ontology=ontology_graph)
with open(gaf_path) as sg:
lines = sum(1 for line in sg)
ttl_path = os.path.join(
os.path.split(gaf_path)[0], "{}_cam.ttl".format(dataset))
click.echo("Producing ttl: {}".format(ttl_path))
rdf_writer = assoc_rdfgen.TurtleRdfWriter()
transformer = assoc_rdfgen.CamRdfTransform(writer=rdf_writer)
parser_config = assocparser.AssocParserConfig(ontology=ontology_graph)
with open(gaf_path) as gf:
with click.progressbar(
iterable=gafparser.association_generator(file=gf),
length=lines) as associations:
for association in associations:
if "header" not in association or not association["header"]:
transformer.provenance()
transformer.translate(association)
with open(ttl_path, "wb") as ttl:
click.echo("Writing ttl to disk")
rdf_writer.serialize(destination=ttl)
return ttl_path
def produce_gpi(dataset, target_dir, gaf_path, ontology_graph):
gafparser = GafParser()
gafparser.config = assocparser.AssocParserConfig(ontology=ontology_graph)
with open(gaf_path) as sg:
lines = sum(1 for line in sg)
gpi_path = os.path.join(
os.path.split(gaf_path)[0], "{}.gpi".format(dataset))
with open(gaf_path) as gf, open(gpi_path, "w") as gpi:
click.echo("Using {} as the gaf to build gpi with".format(gaf_path))
bridge = gafgpibridge.GafGpiBridge()
gpiwriter = entitywriter.GpiWriter(file=gpi)
gpi_cache = set()
with click.progressbar(
iterable=gafparser.association_generator(file=gf),
length=lines) as associations:
for association in associations:
entity = bridge.convert_association(association)
if entity not in gpi_cache and entity is not None:
# If the entity is not in the cache, add it and write it out
gpi_cache.add(entity)
gpiwriter.write_entity(entity)
return gpi_path
def produce_gaf(dataset, source_gaf, ontology_graph, gpipath=None, paint=False, group="unknown"):
filtered_associations = open(os.path.join(os.path.split(source_gaf)[0], "{}_noiea.gaf".format(dataset)), "w")
config = assocparser.AssocParserConfig(
ontology=ontology_graph,
filter_out_evidence=["IEA"],
filtered_evidence_file=filtered_associations,
gpi_authority_path=gpipath,
paint=paint
)
validated_gaf_path = os.path.join(os.path.split(source_gaf)[0], "{}_valid.gaf".format(dataset))
outfile = open(validated_gaf_path, "w")
gafwriter = GafWriter(file=outfile)
click.echo("Validating source GAF: {}".format(source_gaf))
parser = GafParser(config=config, group=group, dataset=dataset)
with open(source_gaf) as sg:
lines = sum(1 for line in sg)
with open(source_gaf) as gaf:
with click.progressbar(iterable=parser.association_generator(file=gaf), length=lines) as associations:
for assoc in associations:
gafwriter.write_assoc(assoc)
outfile.close()
filtered_associations.close()
with open(os.path.join(os.path.split(source_gaf)[0], "{}.report.md".format(dataset)), "w") as report_md:
report_md.write(parser.report.to_markdown())
with open(os.path.join(os.path.split(source_gaf)[0], "{}.report.json".format(dataset)), "w") as report_json:
report_json.write(json.dumps(parser.report.to_report_json(), indent=4))
return [validated_gaf_path, filtered_associations.name]
def make_products(dataset, target_dir, gaf_path, products, ontology_graph):
gafparser = GafParser()
gafparser.config = assocparser.AssocParserConfig(
ontology=ontology_graph,
paint=True
)
with open(gaf_path) as sg:
lines = sum(1 for line in sg)
product_files = {
"gpad": open(os.path.join(os.path.split(gaf_path)[0], "{}.gpad".format(dataset)), "w"),
"ttl": open(os.path.join(os.path.split(gaf_path)[0], "{}_cam.ttl".format(dataset)), "wb")
}
if not products["gpad"] and not products["ttl"]:
# Bail if we have no products
return []
# def write_gpi_entity(association, bridge, gpiwriter):
with open(gaf_path) as gf:
# gpi info:
click.echo("Using {} as the gaf to build data products with".format(gaf_path))
if products["ttl"]:
click.echo("Setting up {}".format(product_files["ttl"].name))
rdf_writer = assoc_rdfgen.TurtleRdfWriter(label=os.path.split(product_files["ttl"].name)[1] )
transformer = assoc_rdfgen.CamRdfTransform(writer=rdf_writer)
parser_config = assocparser.AssocParserConfig(ontology=ontology_graph)
if products["gpad"]:
click.echo("Setting up {}".format(product_files["gpad"].name))
gpadwriter = GpadWriter(file=product_files["gpad"])
click.echo("Making products...")
with click.progressbar(iterable=gafparser.association_generator(file=gf), length=lines) as associations:
for association in associations:
if products["ttl"]:
if "header" not in association or not association["header"]:
transformer.provenance()
transformer.translate(association)
if products["gpad"]:
gpadwriter.write_assoc(association)
# post ttl steps
if products["ttl"]:
click.echo("Writing ttl to disk")
rdf_writer.serialize(destination=product_files["ttl"])
# After we run through associations
for f in product_files.values():
f.close()
return [product_files[prod].name for prod in sorted(product_files.keys()) if products[prod]]
def gafparser_generator(ontology_graph: ontol.Ontology, gaf_file):
config = assocparser.AssocParserConfig(ontology=ontology_graph, )
parser = GafParser(config=config)
return parser.association_generator(gaf_file, skipheader=True)
