def trimming(self, raw_file, outname):
self.raw_file = raw_file
self.outname = outname
default_strand = STRAND
seq_factory_fasta = seq_factory_from_fasta(default_strand)
ltr_seq = seq_factory_fasta.make(self.ltr_fasta)
kwds = {"Program": ncbi_toolkit.EProgram.eBlastn}
ltr_blaster = blaster(ltr_seq, **kwds)
linker_seq = seq_factory_fasta.make(self.linker_fasta)
kwds = {"Program": ncbi_toolkit.EProgram.eBlastn}
linker_blaster = blaster(linker_seq, **kwds)
fin = open(self.raw_file)
fs = fasta_stream_from_stream(fin)
ss = blast_seq_stream(seq_factory_fasta, fs)
bs = blast_result_stream(ltr_blaster, ss)
fbs = similarity_filter(0.89, len(ltr_seq.get_sequence()), bs)
ltr_strs = mapped_stream(cut_ltr_mapper_output_fasta(), fbs)
ltr_seqs = blast_seq_stream(seq_factory_fasta, ltr_strs)
lnk_res = blast_result_stream(linker_blaster, ltr_seqs)
lnk_strs = mapped_stream(cut_linker_mapper_output_fasta(), lnk_res)
for x in lnk_strs:
if len(x[1]) > 19:
line = "\n".join([">" + x[0].replace("|", "_").replace("lcl_", ""), x[1]])
self._setOut(line)
else:
pass
def compute_results(self):
"""
Compute results dict.
"""
self.res_dict = {}
b = blaster(self.sequences, **self.blast_options)
subjects, r = b.blast(self.sequences)
self.assertEqual(len(r), len(self.sequences))
self.assertEqual(subjects, self.sequences)
for hit_list in r:
if hit_list is None:
continue
for hsp_list in hit_list:
query = self.sequences[hsp_list.query_index]
subject = self.sequences[hsp_list.ordinal_id_subject_sequence]
#print "%s vs %s" % (query.id, subject.id)
results = []
for hsp in hsp_list:
res = {}
res['bit_score'] = hsp.bit_score
res['e_value'] = hsp.evalue
(res['query_start'],
res['query_end'],
res['subject_start'],
res['subject_end']) = get_match_endpoints(
hsp.query, hsp.subject, query.length, subject.length,
self.progname
)
results.append(res)
results.sort(key=itemgetter("subject_end"))
results.sort(key=itemgetter("subject_start"))
results.sort(key=itemgetter("query_end"))
results.sort(key=itemgetter("query_start"))
self.res_dict[(query.id, subject.id)] = results
def compute_results(self):
"""
Compute results dict.
"""
self.res_dict = {}
for s1 in self.sequences:
b = blaster(s1, **self.blast_options)
for s2 in self.sequences:
subject, r = b.blast(s2)
self.assertEqual(subject, s2)
hit_list = r[0]
if hit_list is None:
continue
hsp_list = hit_list[0]
results = []
for hsp in hsp_list:
res = {}
res['bit_score'] = hsp.bit_score
res['e_value'] = hsp.evalue
(res['query_start'],
res['query_end'],
res['subject_start'],
res['subject_end']) = get_match_endpoints(
hsp.query, hsp.subject, s1.length, s2.length,
self.progname
)
results.append(res)
results.sort(key=itemgetter("subject_end"))
results.sort(key=itemgetter("subject_start"))
results.sort(key=itemgetter("query_end"))
results.sort(key=itemgetter("query_start"))
self.res_dict[(s1.id, s2.id)] = results
def compute_results(self):
"""
Compute results dict.
"""
self.res_dict = {}
b = blaster(self.sequences, **self.blast_options)
subjects, r = b.blast(self.sequences)
self.assertEqual(len(r), len(self.sequences))
self.assertEqual(subjects, self.sequences)
for hit_list in r:
if hit_list is None:
continue
for hsp_list in hit_list:
query = self.sequences[hsp_list.query_index]
subject = self.sequences[hsp_list.ordinal_id_subject_sequence]
#print "%s vs %s" % (query.id, subject.id)
results = []
for hsp in hsp_list:
res = {}
res['bit_score'] = hsp.bit_score
res['e_value'] = hsp.evalue
(res['query_start'], res['query_end'],
res['subject_start'],
res['subject_end']) = get_match_endpoints(
hsp.query, hsp.subject, query.length, subject.length,
self.progname)
results.append(res)
results.sort(key=itemgetter("subject_end"))
results.sort(key=itemgetter("subject_start"))
results.sort(key=itemgetter("query_end"))
results.sort(key=itemgetter("query_start"))
self.res_dict[(query.id, subject.id)] = results
def compute_results(self):
"""
Compute results dict.
"""
self.res_dict = {}
for s1 in self.sequences:
b = blaster(s1, **self.blast_options)
for s2 in self.sequences:
subject, r = b.blast(s2)
self.assertEqual(subject, s2)
hit_list = r[0]
if hit_list is None:
continue
hsp_list = hit_list[0]
results = []
for hsp in hsp_list:
res = {}
res['bit_score'] = hsp.bit_score
res['e_value'] = hsp.evalue
(res['query_start'], res['query_end'],
res['subject_start'],
res['subject_end']) = get_match_endpoints(
hsp.query, hsp.subject, s1.length, s2.length,
self.progname)
results.append(res)
results.sort(key=itemgetter("subject_end"))
results.sort(key=itemgetter("subject_start"))
results.sort(key=itemgetter("query_end"))
results.sort(key=itemgetter("query_start"))
self.res_dict[(s1.id, s2.id)] = results
def test_set_options(self):
for progname in "blastn", "tblastx":
for seq in self.sequences:
prog = get_program(progname)
self.blaster = blaster(seq, Program=prog)
og = self.blaster.get_options()
os = self.blaster.set_options()
for k in og.keys():
v = og[k]
nv = get_modified_value(v)
os[k] = nv
self.assertEqual(nv, og[k])
def test_set_options(self):
for progname in "blastn", "tblastx":
for seq in self.sequences:
prog = get_program(progname)
self.blaster = blaster(seq, Program=prog)
og = self.blaster.get_options()
os = self.blaster.set_options()
for k in og.keys():
v = og[k]
nv = get_modified_value(v)
os[k] = nv
self.assertEqual(nv, og[k])