def predict_longest_ORFs(seq, min_aa_length):
"""
seq --- should be plain string in all upper case, A/T/C/G
Return all longest ORFs that exceed <min_length>
Returns: dict of <frame> --> list of (flag, <0-based start>, <1-based end>)
NOTE that is the seq is reverse complemented, the handler function needs to rev the coords on its own
"""
start_d, stop_d = ORFscores.find_start_stop_codons(seq)
result = {0: [], 1: [], 2: []}
n, m = len(seq) / 3, len(seq) % 3
for frame in xrange(3):
starts, stops = start_d[frame].keys(), stop_d[frame].keys()
starts.sort()
stops.sort()
#print frame, starts, stops
if len(stops) == 0: # no stop, so just output first (start, last)
if len(starts) > 0 and n - starts[0] + 1 >= min_aa_length:
result[frame].append(('dumb-3partial', starts[0] * 3 + frame,
n * 3 + (frame if frame <= m else 0)))
else: # has stop
if len(starts) == 0: # 5' partial
if stops[0] + 1 >= min_aa_length:
result[frame].append(
('dumb-5partial', frame, stops[0] * 3 + 3 + frame))
else: # has at least one start and one stop
i, j = 0, 0
while j < len(stops):
if i == len(starts): break
if stops[j] - starts[i] + 1 >= min_aa_length:
result[frame].append(
('dumb-complete', starts[i] * 3 + frame,
stops[j] * 3 + 3 + frame))
j += 1 # move stop one step down
while i < len(starts) and starts[i] < stops[j - 1]:
i += 1
# check the very last possible ORF
if i < len(starts) and (
j == len(stops) or
(j < len(stops) and starts[i] > stops[j])
) and n - starts[i] + 1 >= min_aa_length:
result[frame].append(
('dumb-3partial', starts[i] * 3 + frame,
n * 3 + (frame if frame <= m else 0)))
return result
def predict_longest_ORFs(seq, min_aa_length):
"""
seq --- should be plain string in all upper case, A/T/C/G
Return all longest ORFs that exceed <min_length>
Returns: dict of <frame> --> list of (flag, <0-based start>, <1-based end>)
NOTE that is the seq is reverse complemented, the handler function needs to rev the coords on its own
"""
start_d, stop_d = ORFscores.find_start_stop_codons(seq)
result = {0: [], 1: [], 2: []}
n, m = len(seq)/3, len(seq)%3
for frame in xrange(3):
starts, stops = start_d[frame].keys(), stop_d[frame].keys()
starts.sort()
stops.sort()
#print frame, starts, stops
if len(stops) == 0: # no stop, so just output first (start, last)
if len(starts) > 0 and n - starts[0] + 1 >= min_aa_length:
result[frame].append(('dumb-3partial', starts[0]*3+frame, n*3+(frame if frame<=m else 0)))
else: # has stop
if len(starts) == 0: # 5' partial
if stops[0] + 1 >= min_aa_length:
result[frame].append(('dumb-5partial', frame, stops[0]*3+3+frame))
else: # has at least one start and one stop
i, j = 0, 0
while j < len(stops):
if i == len(starts): break
if stops[j] - starts[i] + 1 >= min_aa_length:
result[frame].append(('dumb-complete', starts[i]*3+frame, stops[j]*3+3+frame))
j += 1 # move stop one step down
while i < len(starts) and starts[i] < stops[j-1]:
i += 1
# check the very last possible ORF
if i < len(starts) and (j == len(stops) or (j < len(stops) and starts[i] > stops[j])) and n - starts[i] + 1 >= min_aa_length:
result[frame].append(('dumb-3partial', starts[i]*3+frame, n*3+(frame if frame<=m else 0)))
return result
def predict_longest_ORFs(seq, min_aa_length, use_firstORF=False):
"""
seq --- should be plain string in all upper case, A/T/C/G
Return the longest ORFs that exceed <min_length> (unless use_firstORF is True)
Returns: dict of <frame> --> list of (flag, <0-based start>, <1-based end>)
NOTE that is the seq is reverse complemented, the handler function needs to rev the coords on its own
"""
start_d, stop_d = ORFscores.find_start_stop_codons(seq)
result = {0: [], 1: [], 2: []}
n, m = len(seq) / 3, len(seq) % 3
for frame in xrange(3):
starts, stops = start_d[frame].keys(), stop_d[frame].keys()
starts.sort()
stops.sort()
#print frame, starts, stops
if len(stops) == 0: # no stop, so just output first (start, last)
if len(starts) > 0 and n - starts[0] + 1 >= min_aa_length:
result[frame].append(
('dumb-3partial', starts[0] * 3 + frame,
n * 3 + (frame if frame <= m else frame - 3)))
else: # has stop
if len(starts) == 0: # 5' partial
if stops[0] + 1 >= min_aa_length:
result[frame].append(
('dumb-5partial', frame, stops[0] * 3 + 3 + frame))
else: # has at least one start and one stop
i, j = 0, 0
# if the first stop is smaller than i, find the first j s.t. stops[j-1] < start[0] < stops[j]
if stops[0] < starts[0]:
while j < len(stops) and starts[0] < stops[j - 1]:
j += 1
# now: stops[j-1] < starts[0] < stops[j]
while j < len(stops):
if i == len(starts): break
if stops[j] - starts[i] + 1 >= min_aa_length:
#rint frame, starts[i], stops[j]
result[frame].append(
('dumb-complete', starts[i] * 3 + frame,
stops[j] * 3 + 3 + frame))
j += 1 # move stop one step down
while i < len(starts) and starts[i] < stops[j - 1]:
i += 1
# now starts[i] is between the last stop and this one
# check the very last possible ORF
if i < len(starts) and (
j == len(stops) or
(j < len(stops) and starts[i] > stops[j])
) and n - starts[i] + 1 >= min_aa_length:
result[frame].append(
('dumb-3partial', starts[i] * 3 + frame,
n * 3 + (frame if frame <= m else frame - 3)))
# now pick the frame with the longest ORF!
if all(len(v) == 0 for v in result.itervalues()): # no ORF found
return None
best_frame, best_flag, best_s, best_e, best_len = None, None, None, None, 0
if not use_firstORF: # find the longest ORF among all frames
for _frame, v in result.iteritems():
for (flag, s, e) in v:
_len = e - s
if _len > best_len:
best_frame, best_flag, best_s, best_e, best_len = \
_frame, flag, s, e, _len
else: # use the first ORF among all frames
for _frame, v in result.iteritems():
for (flag, s, e) in v:
_len = e - s
if best_s is None or s < best_s or (s == best_s
and _len > best_len):
best_frame, best_flag, best_s, best_e, best_len = \
_frame, flag, s, e, _len
return {best_frame: [(best_flag, best_s, best_e)]}
def predict_longest_ORFs(seq, min_aa_length, use_firstORF=False):
"""
seq --- should be plain string in all upper case, A/T/C/G
Return the longest ORFs that exceed <min_length> (unless use_firstORF is True)
Returns: dict of <frame> --> list of (flag, <0-based start>, <1-based end>)
NOTE that is the seq is reverse complemented, the handler function needs to rev the coords on its own
"""
start_d, stop_d = ORFscores.find_start_stop_codons(seq)
result = {0: [], 1: [], 2: []}
n, m = len(seq)/3, len(seq)%3
for frame in xrange(3):
starts, stops = start_d[frame].keys(), stop_d[frame].keys()
starts.sort()
stops.sort()
#print frame, starts, stops
if len(stops) == 0: # no stop, so just output first (start, last)
if len(starts) > 0 and n - starts[0] + 1 >= min_aa_length:
result[frame].append(('dumb-3partial', starts[0]*3+frame, n*3+(frame if frame<=m else frame-3)))
else: # has stop
if len(starts) == 0: # 5' partial
if stops[0] + 1 >= min_aa_length:
result[frame].append(('dumb-5partial', frame, stops[0]*3+3+frame))
else: # has at least one start and one stop
i, j = 0, 0
# if the first stop is smaller than i, find the first j s.t. stops[j-1] < start[0] < stops[j]
if stops[0] < starts[0]:
while j < len(stops) and starts[0] < stops[j-1]:
j += 1
# now: stops[j-1] < starts[0] < stops[j]
while j < len(stops):
if i == len(starts): break
if stops[j] - starts[i] + 1 >= min_aa_length:
#rint frame, starts[i], stops[j]
result[frame].append(('dumb-complete', starts[i]*3+frame, stops[j]*3+3+frame))
j += 1 # move stop one step down
while i < len(starts) and starts[i] < stops[j-1]:
i += 1
# now starts[i] is between the last stop and this one
# check the very last possible ORF
if i < len(starts) and (j == len(stops) or (j < len(stops) and starts[i] > stops[j])) and n - starts[i] + 1 >= min_aa_length:
result[frame].append(('dumb-3partial', starts[i]*3+frame, n*3+(frame if frame<=m else frame-3)))
# now pick the frame with the longest ORF!
if all(len(v)==0 for v in result.itervalues()): # no ORF found
return None
best_frame, best_flag, best_s, best_e, best_len = None, None, None, None, 0
if not use_firstORF: # find the longest ORF among all frames
for _frame, v in result.iteritems():
for (flag, s, e) in v:
_len = e - s
if _len > best_len:
best_frame, best_flag, best_s, best_e, best_len = \
_frame, flag, s, e, _len
else: # use the first ORF among all frames
for _frame, v in result.iteritems():
for (flag, s, e) in v:
_len = e - s
if best_s is None or s < best_s or (s==best_s and _len>best_len):
best_frame, best_flag, best_s, best_e, best_len = \
_frame, flag, s, e, _len
return {best_frame: [(best_flag, best_s, best_e)]}
