def get_distances(seq1,
seq2,
per_site=True,
aligned=False,
ignore_gaps=True,
alphabet=None,
aligner_tools=['mafft', 'muscle']):
d = distance(seq1=seq1,
seq2=seq2,
per_site=per_site,
aligned=aligned,
ignore_gaps=ignore_gaps,
alphabet=alphabet,
aligner_tools=aligner_tools)
rc = None
drc = None
if (not alphabet) or (not alphabet.has_state('M')):
try:
rc = sequtils.get_reverse_complement(seq1),
except:
pass
if rc:
drc = distance(seq1=sequtils.get_reverse_complement(seq1),
seq2=seq2,
per_site=per_site,
aligned=False,
ignore_gaps=ignore_gaps,
alphabet=alphabet,
aligner_tools=aligner_tools)
return d, drc
def get_distances(seq1, seq2,
per_site = True,
aligned = False,
ignore_gaps = True,
alphabet = None,
aligner_tools = ['mafft', 'muscle']):
d = distance(
seq1 = seq1,
seq2 = seq2,
per_site = per_site,
aligned = aligned,
ignore_gaps = ignore_gaps,
alphabet = alphabet,
aligner_tools = aligner_tools)
rc = None
drc = None
if (not alphabet) or (not alphabet.has_state('M')):
try:
rc = sequtils.get_reverse_complement(seq1),
except:
pass
if rc:
drc = distance(
seq1 = sequtils.get_reverse_complement(seq1),
seq2 = seq2,
per_site = per_site,
aligned = False,
ignore_gaps = ignore_gaps,
alphabet = alphabet,
aligner_tools = aligner_tools)
return d, drc
def reverse_complement_to_longest_reading_frame(seq_iter,
gap_characters=['-'],
table=1,
allow_partial=True,
require_start_after_stop=True,
log_frequency=0):
for i, s in enumerate(remove_gaps(seq_iter,
gap_characters=gap_characters)):
if (log_frequency > 0) and (((i + 1) % log_frequency) == 0):
_LOG.info('{0}: Checking reverse complement of seq {1}...'.format(
datetime.datetime.now(), (i + 1)))
rc = sequtils.get_reverse_complement(s)
p1 = sequtils.get_longest_reading_frames(
seq_record=s,
table=table,
allow_partial=allow_partial,
require_start_after_stop=require_start_after_stop)
p2 = sequtils.get_longest_reading_frames(
seq_record=rc,
table=table,
allow_partial=allow_partial,
require_start_after_stop=require_start_after_stop)
_LOG.debug('{0}: read length {1}, rev comp read length {2}'.format(
s.id, len(p1[0].seq), len(p2[0].seq)))
if len(p2) == 0:
yield s
elif len(p1) == 0:
_LOG.warning('Reverse complementing sequence {0!r}'.format(rc.id))
yield rc
elif len(p2[0].seq) > len(p1[0].seq):
_LOG.warning('Reverse complementing sequence {0!r}'.format(rc.id))
yield rc
else:
yield s
def reverse_complement_to_first_seq(seq_iter,
per_site=True,
aligned=False,
ignore_gaps=True,
alphabet=None,
aligner_tools=['mafft', 'muscle'],
log_frequency=0):
seq1 = None
for i, seq2 in enumerate(seq_iter):
if i == 0:
seq1 = seq2
yield seq2
continue
if (log_frequency > 0) and (((i + 1) % log_frequency) == 0):
_LOG.info('{0}: Checking reverse complement of seq {1}...'.format(
datetime.datetime.now(), (i + 1)))
d, drc = seqstats.get_distances(seq1=seq1,
seq2=seq2,
per_site=per_site,
aligned=aligned,
ignore_gaps=ignore_gaps,
alphabet=alphabet,
aligner_tools=aligner_tools)
_LOG.debug('{0}: distance {1}, rev comp distance {2}'.format(
seq2.id, d, drc))
if drc < d:
_LOG.warning(
'Reverse complementing sequence {0!r} (length {1})\n\t'
'rev comp distance ({2}) < current distance '
'({3})'.format(seq2.id, len(seq2.seq), drc, d))
yield sequtils.get_reverse_complement(seq2)
continue
yield seq2
def reverse_complement_to_longest_reading_frame(seq_iter,
gap_characters=['-'],
table = 1,
allow_partial = True,
require_start_after_stop = True,
log_frequency = 0):
for i, s in enumerate(remove_gaps(seq_iter, gap_characters=gap_characters)):
if (log_frequency > 0) and (((i + 1) % log_frequency) == 0):
_LOG.info('{0}: Checking reverse complement of seq {1}...'.format(
datetime.datetime.now(),
(i + 1)))
rc = sequtils.get_reverse_complement(s)
p1 = sequtils.get_longest_reading_frames(seq_record = s,
table = table,
allow_partial = allow_partial,
require_start_after_stop = require_start_after_stop)
p2 = sequtils.get_longest_reading_frames(seq_record = rc,
table = table,
allow_partial = allow_partial,
require_start_after_stop = require_start_after_stop)
_LOG.debug('{0}: read length {1}, rev comp read length {2}'.format(
s.id, len(p1[0].seq), len(p2[0].seq)))
if len(p2) == 0:
yield s
elif len(p1) == 0:
_LOG.warning('Reverse complementing sequence {0!r}'.format(rc.id))
yield rc
elif len(p2[0].seq) > len(p1[0].seq):
_LOG.warning('Reverse complementing sequence {0!r}'.format(rc.id))
yield rc
else:
yield s
def reverse_complement_to_first_seq(seq_iter,
per_site = True,
aligned = False,
ignore_gaps = True,
alphabet = None,
aligner_tools = ['mafft', 'muscle'],
log_frequency = 0):
seq1 = None
for i, seq2 in enumerate(seq_iter):
if i == 0:
seq1 = seq2
yield seq2
continue
if (log_frequency > 0) and (((i + 1) % log_frequency) == 0):
_LOG.info('{0}: Checking reverse complement of seq {1}...'.format(
datetime.datetime.now(),
(i + 1)))
d, drc = seqstats.get_distances(
seq1 = seq1,
seq2 = seq2,
per_site = per_site,
aligned = aligned,
ignore_gaps = ignore_gaps,
alphabet = alphabet,
aligner_tools = aligner_tools)
_LOG.debug('{0}: distance {1}, rev comp distance {2}'.format(
seq2.id, d, drc))
if drc < d:
_LOG.warning('Reverse complementing sequence {0!r} (length {1})\n\t'
'rev comp distance ({2}) < current distance '
'({3})'.format(seq2.id, len(seq2.seq), drc, d))
yield sequtils.get_reverse_complement(seq2)
continue
yield seq2
def summarize_longest_read_lengths(seq_iter,
gap_characters=['-'],
table=1,
allow_partial=True,
require_start_after_stop=True):
lengths = []
for seq in seqmod.remove_gaps(seq_iter, gap_characters=gap_characters):
l = 0
rcl = 0
lrf = sequtils.get_longest_reading_frames(
seq,
table=table,
allow_partial=allow_partial,
require_start_after_stop=require_start_after_stop)
if lrf:
l = len(lrf[0].seq)
rc_lrf = sequtils.get_longest_reading_frames(
sequtils.get_reverse_complement(seq),
table=table,
allow_partial=allow_partial,
require_start_after_stop=require_start_after_stop)
if rc_lrf:
rcl = len(rc_lrf[0].seq)
lengths.append((l, rcl, seq.id))
return sorted(lengths)
def summarize_longest_read_lengths(seq_iter,
gap_characters=['-'],
table = 1,
allow_partial = True,
require_start_after_stop = True):
lengths = []
for seq in seqmod.remove_gaps(seq_iter,
gap_characters = gap_characters):
l = 0
rcl = 0
lrf = sequtils.get_longest_reading_frames(
seq,
table = table,
allow_partial = allow_partial,
require_start_after_stop = require_start_after_stop)
if lrf:
l = len(lrf[0].seq)
rc_lrf = sequtils.get_longest_reading_frames(
sequtils.get_reverse_complement(seq),
table = table,
allow_partial = allow_partial,
require_start_after_stop = require_start_after_stop)
if rc_lrf:
rcl = len(rc_lrf[0].seq)
lengths.append((l, rcl, seq.id))
return sorted(lengths)
def reverse_complement(seq_iter):
for s in seq_iter:
yield sequtils.get_reverse_complement(s)
def reverse_complement(seq_iter):
for s in seq_iter:
yield sequtils.get_reverse_complement(s)
