def get_more_fragstats(f5, hascomp, check_time = False):
## f5 is the connection to fast5 file
basecalled = is_basecalled(f5)
input_events = store_input_events(f5, basecalled)
if check_time:
time_error = [int(has_time_error_list(input_events['start']))]
else:
time_error = []
start_time = input_events['start'][0]
end_time = input_events['start'][-1]
nevents = len(input_events['start'])
slope = (end_time-start_time)/(nevents-1)
mean_dur_input = np.mean(input_events['length'])
sd_dur_input = np.std(input_events['length'])
med_dur_input = np.median(input_events['length'])
min_dur_input = min(input_events['length'])
max_dur_input = max(input_events['length'])
template_events = store_template_events(f5, basecalled, getbasecallinfo=True)
tmoves = {0:0, 1:0, 2:0, 3:0, 4:0, 5:0}
for move in template_events['move']:
tmoves[move] += 1
if hascomp:
comp_events = store_complement_events(f5, basecalled, getbasecallinfo=True)
cmoves = {0:0, 1:0, 2:0, 3:0, 4:0, 5:0}
for move in comp_events['move']:
cmoves[move] += 1
else:
cmoves = {0:"-", 1:"-", 2:"-", 3:"-", 4:"-", 5:"-"}
return [start_time, end_time, slope, mean_dur_input, sd_dur_input, med_dur_input, min_dur_input, max_dur_input, tmoves[0], tmoves[1], tmoves[2], tmoves[3], tmoves[4], tmoves[5], cmoves[0], cmoves[1], cmoves[2], cmoves[3], cmoves[4], cmoves[5]] + time_error
def frag_fxn(fast5, args):
fragstats = get_frag_stats(fast5)
if args.quadruplex:
minG, maxN = (int(e) for e in args.g4motif.split(","))
regex = re.compile(get_g4_regex(minG, maxN))
revregex = re.compile(get_g4_revregex(minG, maxN))
if args.extensive:
hascomp = has_complement(fast5.hdf5file)
if args.quadruplex:
fragstats += get_regex_counts_in_fast5(fast5, regex, revregex)
if args.checktime:
fragstats += get_more_fragstats(fast5.hdf5file, hascomp = hascomp, check_time = True)
else:
fragstats += get_more_fragstats(fast5.hdf5file, hascomp = hascomp, check_time = False)
if not args.extensive and args.quadruplex:
fragstats += get_regex_counts_in_fast5(fast5, regex, revregex)
if not args.extensive and args.checktime:
input_events = store_input_events(fast5.hdf5file, basecalled)
fragstats += [int(has_time_error_list(input_events['start']))]
print ("\t").join([str(e) for e in fragstats])
