def test_build_blast_db_from_seqs(self):
"""build_blast_db_from_seqs convenience function works as expected
"""
blast_db, db_files = build_blast_db_from_seqs(self.in_seqs1,
output_dir='/tmp')
self.assertTrue(blast_db.startswith('/tmp/Blast_tmp_db'))
self.assertTrue(blast_db.endswith('.fasta'))
expected_db_files = set([blast_db + ext\
for ext in ['.nhr','.nin','.nsq','.nsd','.nsi','.log']])
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(\
len(blastn(self.test_seq,blast_db=blast_db)),1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))
def test_blast_against_new_db(self):
"""Formatdb: blastall against a newly created DB functions as expected
"""
fdb = FormatDb(WorkingDir="/tmp")
result = fdb(self.in_seqs1_fp)
blast_res = blastn(self.test_seq, blast_db=self.in_seqs1_fp)
result.cleanUp()
# Test that a blast result was returned
self.assertTrue("s1" in blast_res, "Not getting any blast results.")
# Test that the sequence we expect was a good blast hit
subject_ids = [r["SUBJECT ID"] for r in blast_res["s1"][0]]
self.assertTrue("11472384" in subject_ids, "Not getting expected blast results.")
def test_blast_against_new_db(self):
"""Formatdb: blastall against a newly created DB functions as expected
"""
fdb = FormatDb(WorkingDir='/tmp')
result = fdb(self.in_seqs1_fp)
blast_res = blastn(self.test_seq, blast_db=self.in_seqs1_fp)
result.cleanUp()
# Test that a blast result was returned
self.assertTrue('s1' in blast_res,\
"Not getting any blast results.")
# Test that the sequence we expect was a good blast hit
subject_ids = [r['SUBJECT ID'] for r in blast_res['s1'][0]]
self.assertTrue('11472384' in subject_ids,\
"Not getting expected blast results.")
def test_build_blast_db_from_fasta_path_aln(self):
"""build_blast_db_from_fasta_path works with alignment as input
"""
blast_db, db_files = build_blast_db_from_fasta_path(self.in_aln1_fp)
self.assertEqual(blast_db, self.in_aln1_fp)
expected_db_files = set([blast_db + ext for ext in [".nhr", ".nin", ".nsq", ".nsd", ".nsi", ".log"]])
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(len(blastn(self.test_seq, blast_db=blast_db, e_value=0.0)), 1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))
def test_build_blast_db_from_fasta_path(self):
"""build_blast_db_from_fasta_path convenience function works as expected
"""
blast_db, db_files = build_blast_db_from_fasta_path(self.in_seqs1_fp)
self.assertEqual(blast_db, self.in_seqs1_fp)
expected_db_files = set([self.in_seqs1_fp + ext for ext in [".nhr", ".nin", ".nsq", ".nsd", ".nsi", ".log"]])
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(len(blastn(self.test_seq, blast_db=blast_db)), 1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))
def test_build_blast_db_from_fasta_path_aln(self):
"""build_blast_db_from_fasta_path works with alignment as input
"""
blast_db, db_files = build_blast_db_from_fasta_path(self.in_aln1_fp)
self.assertEqual(blast_db, self.in_aln1_fp)
expected_db_files = set([blast_db + ext\
for ext in ['.nhr','.nin','.nsq','.nsd','.nsi','.log']])
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(\
len(blastn(self.test_seq,blast_db=blast_db,e_value=0.0)),1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))
def test_build_blast_db_from_seqs(self):
"""build_blast_db_from_seqs convenience function works as expected
"""
blast_db, db_files = build_blast_db_from_seqs(self.in_seqs1, output_dir="/tmp")
self.assertTrue(blast_db.startswith("/tmp/Blast_tmp_db"))
self.assertTrue(blast_db.endswith(".fasta"))
expected_db_files = set([blast_db + ext for ext in [".nhr", ".nin", ".nsq", ".nsd", ".nsi", ".log"]])
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(len(blastn(self.test_seq, blast_db=blast_db)), 1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))
def test_build_blast_db_from_fasta_file(self):
"""build_blast_db_from_fasta_file works with open files as input
"""
blast_db, db_files = build_blast_db_from_fasta_file(open(self.in_aln1_fp), output_dir="/tmp/")
self.assertTrue(blast_db.startswith("/tmp/BLAST_temp_db"))
self.assertTrue(blast_db.endswith(".fasta"))
expected_db_files = set(
[blast_db] + [blast_db + ext for ext in [".nhr", ".nin", ".nsq", ".nsd", ".nsi", ".log"]]
)
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(len(blastn(self.test_seq, blast_db=blast_db, e_value=0.0)), 1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))
def pick_parents(blastdb, ref_seqs,target_sequences,nhits=10,params=None):
"""Pick the most likely parents for a chimeric sequence
blastdb : pre-formatted db of all parents
ref_seqs : aligned parent sequences
target_sequences : a dict of sequences
nhits : number of "best" parents to take
"""
for seq_id, seq in target_sequences:
query_seqs = [("%d" % i, s) for i,s in \
enumerate(fractionate_sequence(seq))]
blastdb = os.path.join(os.getcwd(), blastdb)
result = blastn(query_seqs, blastdb)
ref_db = {}
for q, best_hits in result.bestHitsByQuery(n=nhits):
for rec in best_hits:
id_ = rec['SUBJECT ID']
ref_db[rec['SUBJECT ID']] = ref_seqs[id_]
if not ref_db:
continue
yield (ref_db,seq_id, seq)
def pick_parents(blastdb, ref_seqs, target_sequences, nhits=10, params=None):
"""Pick the most likely parents for a chimeric sequence
blastdb : pre-formatted db of all parents
ref_seqs : aligned parent sequences
target_sequences : a dict of sequences
nhits : number of "best" parents to take
"""
for seq_id, seq in target_sequences:
query_seqs = [("%d" % i, s) for i,s in \
enumerate(fractionate_sequence(seq))]
blastdb = os.path.join(os.getcwd(), blastdb)
result = blastn(query_seqs, blastdb)
ref_db = {}
for q, best_hits in result.bestHitsByQuery(n=nhits):
for rec in best_hits:
id_ = rec['SUBJECT ID']
ref_db[rec['SUBJECT ID']] = ref_seqs[id_]
if not ref_db:
continue
yield (ref_db, seq_id, seq)
def test_build_blast_db_from_fasta_file(self):
"""build_blast_db_from_fasta_file works with open files as input
"""
blast_db, db_files = \
build_blast_db_from_fasta_file(open(self.in_aln1_fp),output_dir='/tmp/')
self.assertTrue(blast_db.startswith('/tmp/BLAST_temp_db'))
self.assertTrue(blast_db.endswith('.fasta'))
expected_db_files = set([blast_db] + [blast_db + ext\
for ext in ['.nhr','.nin','.nsq','.nsd','.nsi','.log']])
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(\
len(blastn(self.test_seq,blast_db=blast_db,e_value=0.0)),1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))
def test_build_blast_db_from_fasta_path(self):
"""build_blast_db_from_fasta_path convenience function works as expected
"""
blast_db, db_files = \
build_blast_db_from_fasta_path(self.in_seqs1_fp)
self.assertEqual(blast_db, self.in_seqs1_fp)
expected_db_files = set([self.in_seqs1_fp + ext\
for ext in ['.nhr','.nin','.nsq','.nsd','.nsi','.log']])
self.assertEqual(set(db_files), expected_db_files)
# result returned when blasting against new db
self.assertEqual(\
len(blastn(self.test_seq,blast_db=blast_db)),1)
# Make sure all db_files exist
for fp in db_files:
self.assertTrue(exists(fp))
# Remove all db_files exist
remove_files(db_files)
# Make sure nothing weird happened in the remove
for fp in db_files:
self.assertFalse(exists(fp))