def UnknownBases(transcript_dict, seq_dict):
r = re.compile("[atgcATGC][N]+[atgcATGC]")
classify_dict = {}
details_dict = {}
for ens_id, t in transcript_iterator(transcript_dict):
s = t.get_mrna(seq_dict)
tmp = [seq_lib.transcript_coordinate_to_bed(t, m.start() + 1, m.end() - 1, rgb, sys._getframe().f_code.co_name)
for m in re.finditer(r, s)]
if len(tmp) > 0:
details_dict[ens_id] = tmp
classify_dict[ens_id] = 1
else:
classify_dict[ens_id] = 0
return classify_dict, details_dict
def unknown_base(transcript_dict, seq_dict, r, cds):
classify_dict = {}
details_dict = {}
for ens_id, t in transcript_iterator(transcript_dict):
if cds is True:
s = t.get_cds(seq_dict)
tmp = [seq_lib.cds_coordinate_to_bed(t, m.start(), m.end(), rgb, sys._getframe().f_code.co_name) for m in
re.finditer(r, s)]
else:
s = t.get_mrna(seq_dict)
tmp = [seq_lib.transcript_coordinate_to_bed(t, m.start(), m.end(), rgb, sys._getframe().f_code.co_name)
for m in re.finditer(r, s)]
if len(tmp) > 0:
details_dict[ens_id] = tmp
classify_dict[ens_id] = 1
else:
classify_dict[ens_id] = 0
return classify_dict, details_dict
