def setUp(self):
if not testutil.blast_enabled():
raise SkipTest, "no BLAST installed"
hbb1_mouse = testutil.datafile('hbb1_mouse.fa')
sp_hbb1 = testutil.datafile('sp_hbb1')
self.dna = seqdb.SequenceFileDB(hbb1_mouse)
self.prot = seqdb.SequenceFileDB(sp_hbb1)
def test_blastp(self):
"Testing blastp"
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastMapping(self.prot, verbose=False)
results = blastmap[self.prot["HBB1_XENLA"]]
check_results_relaxed_blastp([results], blastp_correct_results, pair_identity_tuple, allowedLengthDiff=2)
def test_seq_without_db(self):
"Check that sequences without associated DBs work as query strings"
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastMapping(self.prot, verbose=False)
seq = self.prot["HBB1_XENLA"]
seq_no_db = sequence.Sequence(str(seq), "HBB1_XENLA_no_db")
slice = blastmap(seq=seq_no_db)[seq_no_db]
assert len(slice)
def test_tblastn_no_blastx(self):
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastMapping(self.prot)
try:
results = blastmap[self.dna["gi|171854975|dbj|AB364477.1|"]]
raise AssertionError("failed to trap blastx in BlastMapping")
except ValueError:
pass
def test_blastx_no_blastp(self):
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastxMapping(self.prot, verbose=False)
try:
results = blastmap(self.prot["HBB1_MOUSE"])
raise AssertionError("failed to trap blastp in BlastxMapping")
except ValueError:
pass
def test_blastx_rc(self):
"Testing blastx with negative frames"
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastxMapping(self.prot, verbose=False)
correct = [
(143, 143, 429, 0.53146853146853146),
(143, 145, 429, 0.28275862068965518),
(143, 145, 429, 0.28965517241379313),
(143, 145, 429, 0.29655172413793102),
(143, 145, 429, 0.30344827586206896),
(144, 144, 432, 0.4513888888888889),
(144, 144, 432, 0.4513888888888889),
(145, 145, 435, 0.45517241379310347),
(145, 145, 435, 0.51034482758620692),
(146, 142, 438, 0.35616438356164382),
(146, 146, 438, 0.4589041095890411),
(146, 146, 438, 0.46575342465753422),
(146, 146, 438, 0.4726027397260274),
(146, 146, 438, 0.4726027397260274),
(146, 146, 438, 0.4863013698630137),
(146, 146, 438, 0.59589041095890416),
(146, 146, 438, 0.62328767123287676),
(146, 146, 438, 0.66438356164383561),
(146, 146, 438, 0.74657534246575341),
(146, 146, 438, 0.91095890410958902),
(146, 146, 438, 0.97945205479452058),
]
results = blastmap[self.dna_rc["hbb1_mouse_RC"]]
check_results_relaxed_blastx(
results,
correct,
lambda t: (len(t[0]), len(t[1]), len(t[0].sequence), t[2].pIdentity()),
allowedLengthDiff=2,
)
results = blastmap[self.dna_rc["hbb1_mouse_RC_2"]]
check_results_relaxed_blastx(
results,
correct,
lambda t: (len(t[0]), len(t[1]), len(t[0].sequence), t[2].pIdentity()),
allowedLengthDiff=2,
)
results = blastmap[self.dna_rc["hbb1_mouse_RC_3"]]
check_results_relaxed_blastx(
results,
correct,
lambda t: (len(t[0]), len(t[1]), len(t[0].sequence), t[2].pIdentity()),
allowedLengthDiff=2,
)
def test_tblastn(self):
"tblastn test"
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastMapping(self.dna, verbose=False)
correct = [(144, 144, 432, 0.451)]
result = blastmap[self.prot["HBB1_XENLA"]]
check_results_relaxed_blastx(
[result], correct, lambda t: (len(t[1]), len(t[0]), len(t[1].sequence), t[2].pIdentity())
)
def test_translation_db_in_results_of_db_search(self):
"""
Test that the NLMSA in a BlastxMapping properly picks up the
translationDB from the query sequence dict.
"""
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastxMapping(self.prot, verbose=False)
results = blastmap(queryDB=self.dna)
tdb = translationDB.get_translation_db(self.dna)
assert tdb.annodb in results.seqDict.dicts
def test_multiblast_long(self):
"testing multi sequence blast with long db"
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
longerFile = testutil.datafile("sp_all_hbb")
sp_all_hbb = seqdb.SequenceFileDB(longerFile)
blastmap = blast.BlastMapping(self.prot, verbose=False)
al = cnestedlist.NLMSA("blasthits", "memory", pairwiseMode=True, bidirectional=False)
blastmap(None, al, queryDB=sp_all_hbb) # all vs all
al.build() # construct the alignment indexes
def test_non_consumable_results(self):
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastxMapping(self.prot, verbose=False)
query_seq = self.dna["gi|171854975|dbj|AB364477.1|"]
results = blastmap[query_seq]
x = list(results)
y = list(results)
assert len(x), x
assert x == y, "BlastxMapping.__getitem__ should return list"
def test_megablast(self):
"""test megablast"""
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.MegablastMapping(self.dna, verbose=False)
# must use copy of sequence to get "self matches" from NLMSA...
query = seqdb.Sequence(str(self.dna["gi|171854975|dbj|AB364477.1|"]), "foo")
try:
result = blastmap[query]
except OSError: # silently ignore missing RepeatMasker, megablast
return
found = [(len(t[0]), len(t[1])) for t in result.edges()]
assert found == [(444, 444)]
def test_translation_db_in_results_of_seq_search(self):
"""
Test that the NLMSA in a BlastxMapping properly picks up the
translationDB from a single input sequence.
"""
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastxMapping(self.prot, verbose=False)
query_seq = self.dna["gi|171854975|dbj|AB364477.1|"]
results = blastmap(seq=query_seq)
tdb = translationDB.get_translation_db(self.dna)
assert tdb.annodb in results.seqDict.dicts
def test_maskEnd(self):
"""
This tests against a minor bug in cnestedlist where maskEnd
is used to clip the end to the mask region.
"""
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
db = self.gapping
blastmap = blast.BlastMapping(db)
ungapped = db["ungapped"]
gapped = db["gapped"]
results = blastmap[gapped]
results[ungapped]
def test_multiblast_single(self):
"Test multi-sequence BLAST results, for BLASTs run one by one."
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastMapping(self.prot, verbose=False)
al = cnestedlist.NLMSA("blasthits", "memory", pairwiseMode=True, bidirectional=False)
for seq in self.prot.values():
blastmap(seq, al) # all vs all, one by one
al.build() # construct the alignment indexes
results = [al[seq] for seq in self.prot.values()]
results_multi = self.get_multiblast_results()
# Strict check must work here even on live BLAST output
check_results(results, results_multi, pair_identity_tuple)
def test_translated_seqs_in_results(self):
"""
Only NLMSASlices for the query sequence should show up in
BlastxMapping.__getitem__, right?
"""
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
blastmap = blast.BlastxMapping(self.prot, verbose=False)
query_seq = self.dna["gi|171854975|dbj|AB364477.1|"]
results = blastmap[query_seq]
tdb = translationDB.get_translation_db(self.dna)
annodb = tdb.annodb
for slice in results:
assert slice.seq.id in annodb, "%s not in annodb!" % slice.seq.id
def test_no_bidirectional(self):
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
db = self.gapping
gapped = db["gapped"]
ungapped = db["ungapped"]
blastmap = blast.BlastMapping(db)
al = blastmap(queryDB=db)
slice = al[gapped]
found_once = False
for src, dest, edge in al[gapped].edges():
if src == gapped[0:40] and dest == ungapped[0:40]:
assert not found_once, "BLAST results should not be bidirectional"
found_once = True
assert found_once, "should have found this match exactly once!"
def get_multiblast_results(self):
"""return saved results or generate them if needed;
results are saved so we only do this time-consuming operation once"""
global _multiblast_results
if not testutil.blast_enabled():
raise SkipTest("no BLAST installed")
if not _multiblast_results:
logger.info("running expensive multiblast")
blastmap = blast.BlastMapping(self.prot, verbose=False)
al = cnestedlist.NLMSA("blasthits", "memory", pairwiseMode=True, bidirectional=False)
blastmap(al=al, queryDB=self.prot) # all vs all
al.build() # construct the alignment indexes
results = [al[seq] for seq in self.prot.values()]
_multiblast_results = reformat_results(results, pair_identity_tuple)
return _multiblast_results