def test_gpi_iterator(self):
"""Test GOA GPI file iterator, gpi-version: 1.1."""
recs = []
with open("UniProt/gp_information.goa_yeast.28.gpi") as handle:
for rec in GOA.gpi_iterator(handle):
recs.append(rec)
self.assertEqual(len(recs), 300)
self.assertEqual(sorted(recs[0].keys()), sorted(GOA.GPI11FIELDS))
# Check values of first record
self.assertEqual(recs[0]["DB_Object_ID"], "A2P2R3")
self.assertEqual(recs[0]["DB_Object_Symbol"], "YMR084W")
self.assertEqual(
recs[0]["DB_Object_Name"],
[
"Putative glutamine--fructose"
"-6-phosphate aminotransferase"
" [isomerizing]"
],
)
self.assertEqual(recs[0]["DB_Object_Synonym"],
["YM084_YEAST", "YMR084W"])
self.assertEqual(recs[0]["DB_Object_Type"], "protein")
self.assertEqual(recs[0]["Taxon"], "taxon:559292")
self.assertEqual(recs[0]["Parent_Object_ID"], "")
self.assertEqual(recs[0]["DB_Xref"], [""])
self.assertEqual(recs[0]["Gene_Product_Properties"],
["db_subset=Swiss-Prot"])
def test_gpi_iterator_one_two(self):
"""Test GOA GPI file iterator, gpi-version: 1.2."""
recs = []
with open("UniProt/goa_human_sample.gpi") as handle:
for rec in GOA.gpi_iterator(handle):
recs.append(rec)
self.assertEqual(len(recs), 9)
self.assertEqual(sorted(recs[0].keys()), sorted(GOA.GPI12FIELDS))
# Check values of first record
self.assertEqual(recs[0]["DB"], "UniProtKB")
self.assertEqual(recs[0]["DB_Object_ID"], "A0A024R1R8")
self.assertEqual(recs[0]["DB_Object_Symbol"], "hCG_2014768")
self.assertEqual(recs[0]["DB_Object_Name"],
["HCG2014768, isoform CRA_a"])
self.assertEqual(recs[0]["DB_Object_Synonym"], ["hCG_2014768"])
self.assertEqual(recs[0]["DB_Object_Type"], "protein")
self.assertEqual(recs[0]["Taxon"], "taxon:9606")
self.assertEqual(recs[0]["Parent_Object_ID"], "")
self.assertEqual(recs[0]["DB_Xref"], [""])
self.assertEqual(recs[0]["Gene_Product_Properties"],
["db_subset=TrEMBL"])
def parse_gpi(infile, taxon=''):
sp_id = defaultdict()
infile_handle = open(infile, 'r')
parser = GOAParser.gpi_iterator(infile_handle)
for rec in parser:
print rec.keys()
if not rec.has_key('Gene_Product_Properties'):
print "This version of the gp information file does not contain all required information"
sys.exit(1)
else:
break
for rec in parser:
taxid = rec['Taxon'].split(':')[1].strip()
db = rec['Gene_Product_Properties'][0].split('=')[1].strip()
if db.startswith('Swiss-Prot') and taxon == taxid:
sp_id[rec['DB_Object_ID']] = 1
return sp_id
def test_gpi_iterator(self):
"""Test GOA GPI file iterator."""
recs = []
with open('UniProt/gp_information.goa_yeast.28.gpi', 'r') as handle:
for rec in GOA.gpi_iterator(handle):
recs.append(rec)
self.assertEqual(len(recs), 300)
self.assertEqual(sorted(recs[0].keys()), sorted(GOA.GPI11FIELDS))
# Check values of first record
self.assertEqual(recs[0]['DB_Object_ID'], 'A2P2R3')
self.assertEqual(recs[0]['DB_Object_Symbol'], 'YMR084W')
self.assertEqual(recs[0]['DB_Object_Name'], [
'Putative glutamine--fructose'
'-6-phosphate aminotransferase'
' [isomerizing]'
])
self.assertEqual(recs[0]['DB_Object_Synonym'],
['YM084_YEAST', 'YMR084W'])
self.assertEqual(recs[0]['DB_Object_Type'], 'protein')
self.assertEqual(recs[0]['Taxon'], 'taxon:559292')
self.assertEqual(recs[0]['Parent_Object_ID'], '')
self.assertEqual(recs[0]['DB_Xref'], [''])
self.assertEqual(recs[0]['Gene_Product_Properties'],
['db_subset=Swiss-Prot'])