def test_header_newline():
gpi_obj = {
'id': "MGI:MGI:1918911",
'label': "0610005C13Rik", # db_object_symbol,
'full_name': "RIKEN cDNA 0610005C13 gene", # db_object_name,
'synonyms': [],
'type': ["gene"], # db_object_type,
'parents': "", # GAF does not have this field, but it's optional in GPI
'xrefs': "", # GAF does not have this field, but it's optional in GPI
'taxon': {
'id': "NCBITaxon:10090"
}
}
entity = gafgpibridge.Entity(gpi_obj)
out = io.StringIO()
gpiwriter = entitywriter.GpiWriter(file=out)
gpiwriter.write_entity(entity)
outlines = out.getvalue().split("\n")
expected_lines = [
"!gpi-version: 1.2",
"MGI\tMGI:1918911\t0610005C13Rik\tRIKEN cDNA 0610005C13 gene\t\tgene\ttaxon:10090\t\t\t",
""
]
assert expected_lines == outlines
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