def test_add_feature_to_record(self):
from barkley import seqrecs
from barkley import blast
blast.make_blast_db(in_file_fasta=TARGET_FILE, db_name=BLAST_DB_NAME, out_dir=TEST_DATA_PATH)
blast.run_blast(
queries_file=QUERIES_FILE,
results_file=BLAST_RESULTS_FILE,
db=BLAST_DB_NAME,
evalue=1e-1,
program="blastn",
task=None,
blast_db_dir=TEST_DATA_PATH,
output_format=5,
num_threads=1,
)
blast_hits = blast.parse_blast_results_xml(file_name=BLAST_RESULTS_FILE)
record = seqrecs.read_sequence_file(
file_path=TARGET_FILE,
file_format="fasta",
ret_type="list",
key="id", # id gi description
alphabet="dna", # dna aa rna
)[0]
hsps = blast_hits["query_01"][0]
for hsp in hsps:
seqrecs.add_feature_to_record(
record=record,
f_start=hsp["sbjct_start"],
f_end=hsp["sbjct_end"],
f_strand=hsp["strand"],
f_type="BLAST_HIT",
f_qualifiers_dict=hsp,
)
# seqrecs.write_sequence_file(
# records=[record],
# file_path=TARGET_FILE_GB,
# file_format='genbank')
self.assertEqual(len(record.features), 1)
def test_run_blast(self):
from barkley import blast
blast.make_blast_db(in_file_fasta=TARGET_FILE, db_name=BLAST_DB_NAME, out_dir=TEST_DATA_PATH)
blast.run_blast(
queries_file=QUERIES_FILE,
results_file=BLAST_RESULTS_FILE,
db=BLAST_DB_NAME,
evalue=1e-1,
program="blastn",
task=None,
blast_db_dir=TEST_DATA_PATH,
output_format=5,
num_threads=1,
)
self.assertTrue(os.path.exists(BLAST_RESULTS_FILE))
def test_parse_blast_results_xml(self):
from barkley import blast
blast.make_blast_db(in_file_fasta=TARGET_FILE, db_name=BLAST_DB_NAME, out_dir=TEST_DATA_PATH)
blast.run_blast(
queries_file=QUERIES_FILE,
results_file=BLAST_RESULTS_FILE,
db=BLAST_DB_NAME,
evalue=1e-1,
program="blastn",
task=None,
blast_db_dir=TEST_DATA_PATH,
output_format=5,
num_threads=1,
)
return_value = blast.parse_blast_results_xml(file_name=BLAST_RESULTS_FILE)
hsp_dict = {
"sbjct_end": 351,
"target_name": "target_01",
"query_start": 1,
"query_end": 280,
"query_name": "query_01",
"expect": 2.48939e-151,
"sbjct_start": 629,
"strand": -1,
}
self.assertEqual(len(return_value.keys()), 2)
self.assertEqual(return_value["query_01"][0][0]["query_name"], hsp_dict["query_name"])
self.assertEqual(return_value["query_01"][0][0]["target_name"], hsp_dict["target_name"])
self.assertEqual(return_value["query_01"][0][0]["sbjct_start"], hsp_dict["sbjct_start"])
self.assertEqual(return_value["query_01"][0][0]["sbjct_end"], hsp_dict["sbjct_end"])
self.assertEqual(return_value["query_01"][0][0]["query_start"], hsp_dict["query_start"])
self.assertEqual(return_value["query_01"][0][0]["query_end"], hsp_dict["query_end"])
self.assertEqual(return_value["query_01"][0][0]["strand"], hsp_dict["strand"])