def findSequencePositions(start_position, end_position, start_sequence,
end_sequence, sentinel, headers, seqs):
"""Find sequence positions. If start_position is a number, it is returned. Otherwise, start_sequence is searched for."""
res_start_position = None
res_end_position = None
def found_sentinel(x):
return sentinel is None or (sentinel in x)
if not start_position is None:
res_start_position = start_position
else:
if not start_sequence is None:
# Find this sequence
for (h, s) in zip(headers, seqs):
if found_sentinel(h):
ind = geneutil.gappedFind(s, start_sequence, start=True)
if ind > 0:
res_start_position = ind
break
if not end_position is None:
end_position = end_position
else:
if not end_sequence is None:
# Find this sequence
for (h, s) in zip(headers, seqs):
if found_sentinel(h):
endpos = geneutil.gappedFind(s, end_sequence, start=False)
if endpos > 0:
res_end_position = endpos
break
return res_start_position, res_end_position
def findSequencePositions(start_position, end_position, start_sequence, end_sequence, sentinel, headers, seqs): """Find sequence positions. If start_position is a number, it is returned. Otherwise, start_sequence is searched for.""" res_start_position = None res_end_position = None def found_sentinel(x): return sentinel is None or (sentinel in x) if not start_position is None: res_start_position = start_position else: if not start_sequence is None: # Find this sequence for (h,s) in zip(headers,seqs): if found_sentinel(h): ind = geneutil.gappedFind(s, start_sequence, start=True) if ind > 0: res_start_position = ind break if not end_position is None: end_position = end_position else: if not end_sequence is None: # Find this sequence for (h,s) in zip(headers,seqs): if found_sentinel(h): endpos = geneutil.gappedFind(s, end_sequence, start=False) if endpos > 0: res_end_position = endpos break return res_start_position, res_end_position
def test_gapped_find(self):
"""Gapped-find testcases"""
self.assertTrue(geneutil.gappedFind('AAASS--SAA','SSS')==3)
self.assertTrue(geneutil.gappedFind('AAASS--SAA','SSSS')==-1)
#print geneutil.gappedFind('AAASS--SAA','SSS',start=False)
self.assertTrue(geneutil.gappedFind('AAASS--SAA','SSS',start=False)==8)
self.assertTrue(geneutil.gappedFind('AAASSxxSAA','SSS',start=False)==-1)
self.assertTrue(geneutil.gappedFind('AAASSxxSAA','SSS',start=False,gap='x')==8)
def test_gapped_find(self):
"""Gapped-find testcases"""
self.assertTrue(geneutil.gappedFind('AAASS--SAA', 'SSS') == 3)
self.assertTrue(geneutil.gappedFind('AAASS--SAA', 'SSSS') == -1)
#print geneutil.gappedFind('AAASS--SAA','SSS',start=False)
self.assertTrue(
geneutil.gappedFind('AAASS--SAA', 'SSS', start=False) == 8)
self.assertTrue(
geneutil.gappedFind('AAASSxxSAA', 'SSS', start=False) == -1)
self.assertTrue(
geneutil.gappedFind('AAASSxxSAA', 'SSS', start=False, gap='x') ==
8)
if options.query in h:
query_ids.append(xi)
if len(query_ids) == 0:
info_outs.write("# Could not find sequences '{}'; exiting\n".format(options.query))
sys.exit()
if len(query_ids) > 1:
info_outs.write("# Found more than one sequence matching '{}'; using the first one: \n#\t{}\n".format(options.query, headers[xi]))
# Pick the first one
query_id = query_ids[0]
gap = '-'
# Find subsequence
(start_position,end_position) = (options.start_position, options.end_position)
if not options.start_sequence is None:
assert not options.end_sequence is None
start_index = geneutil.gappedFind(seqs[query_id], options.start_sequence, gapless=False, gap=gap)
end_index = geneutil.gappedFind(seqs[query_id], options.end_sequence, start=False, gapless=False, gap=gap)
#print(start_index, end_index)
#print(seqs[query_id][start_index:end_index])
elif not options.start_position is None:
assert not options.end_position is None
start_index = geneutil.gappedIndex(seqs[query_id], options.start_position, gap=gap)
end_index = geneutil.gappedIndex(seqs[query_id], options.end_position, gap=gap)
else: # No start/end given;
info_outs.write("# No starting position or sequence given; nothing to do. Exiting\n")
new_headers = []
new_seqs = []
for (h,seq) in zip(headers,seqs):
if not options.exclude:
ex_seq = seq[start_index:end_index]
sys.exit()
if len(query_ids) > 1:
info_outs.write(
"# Found more than one sequence matching '{}'; using the first one: \n#\t{}\n"
.format(options.query, headers[xi]))
# Pick the first one
query_id = query_ids[0]
gap = '-'
# Find subsequence
(start_position, end_position) = (options.start_position,
options.end_position)
if not options.start_sequence is None:
assert not options.end_sequence is None
start_index = geneutil.gappedFind(seqs[query_id],
options.start_sequence,
gapless=False,
gap=gap)
end_index = geneutil.gappedFind(seqs[query_id],
options.end_sequence,
start=False,
gapless=False,
gap=gap)
print(start_index, end_index)
print(seqs[query_id][start_index:end_index])
elif not options.start_position is None:
assert not options.end_position is None
start_index = geneutil.gappedIndex(seqs[query_id],
options.start_position,
gap=gap)
end_index = geneutil.gappedIndex(seqs[query_id],
options.end_position,
