def process_hdf5(arg):
# Extract info from arguments
filename, seqids, threedarray, procampres, seqlen, args = arg
# Open hdf5 file
hdf = h5py.File(filename, 'r')
# Process each read in the hdf file
for read in hdf['Analyses']['EventDetection_000']['Reads']:
events = hdf['Analyses']['EventDetection_000']['Reads'][read][
'Events'][()]
event_collection = list()
for event in events:
event_collection.append(float(event[0]))
# We ignore the first 50 events (Protein driver) and process the following 250 events
squiggle = event_collection[50:300]
# Search squiggle in reference squiggle
squiggleres = squiggle_search2(squiggle, 0, 0, args, seqids,
threedarray, seqlen)
if True:
# logger.info("[%s, %s, %s]", squiggleres[0], squiggleres[2], squiggleres[3])
try:
result = go_or_no(squiggleres[0], squiggleres[2],
squiggleres[3], seqlen, args)
except Exception as err:
logger.error("%s", err)
hdf.close()
return (result, filename, squiggleres)
def process_hdf5((filename, seqids, threedarray, procampres, seqlen, args)):
returndict = dict()
hdf = h5py.File(filename, 'r')
for read in hdf['Analyses']['EventDetection_000']['Reads']:
events = hdf['Analyses']['EventDetection_000']['Reads'][read][
'Events'][()]
event_collection = list()
for event in events:
event_collection.append(float(event[0]))
squiggle = event_collection[50:300]
#print seqlen
#squiggle,channel_id,read_id,args,seqids,threedarray,seqlen
squiggleres = squiggle_search2(squiggle, 0, 0, args, seqids,
threedarray, seqlen)
if 1:
print squiggleres[0], squiggleres[2], squiggleres[3]
try:
#seqid,direction,position,seqlen,args,distance,refmatchlen,querymatchlen
#print squiggleres
#result = go_or_no2(squiggleres[0],squiggleres[2],squiggleres[3],seqlen,args,squiggleres[1],0,0)
result = go_or_no(squiggleres[0], squiggleres[2],
squiggleres[3], seqlen, args)
except Exception, err:
print "ERROR"
print err
def mp_worker((channel_id, data, kmerhash, seqlen, readstarttime,
kmerhash_subset)):
if ((time.time() - readstarttime) > args.time):
print "We have a timeout"
return 'timeout', channel_id, data.read_id, data.events[0].start
else:
try:
print "Read start time", readstarttime
squiggle = extractsquig(data.events)
squiggleres = squiggle_search2(squiggle, channel_id, data.read_id,
kmerhash, seqlen)
print "Full Length:", squiggleres
print "Full Length Match Length:", squiggleres[5] - squiggleres[3]
squiggleres2 = squiggle_search2(squiggle, channel_id, data.read_id,
kmerhash_subset, seqlen)
squiggleres3 = squiggle_search2(squiggle[100:200], channel_id,
data.read_id, kmerhash_subset,
seqlen)
squiggleres4 = squiggle_search2(squiggle[50:100], channel_id,
data.read_id, kmerhash_subset,
seqlen)
if squiggleres2[5] > squiggleres3[3] > squiggleres2[
3] and squiggleres3[3] > squiggleres4[3] > squiggleres2[3]:
print "!!!!!!!!!!!!!!!! We got a good one! !!!!!!!!!!!!!!!!"
print "Subset:", squiggleres2
print "Subset Match Length:", squiggleres2[5] - squiggleres2[3]
print "SecondHalf:", squiggleres3
print "SecondHalf:", squiggleres3[5] - squiggleres3[3]
print "FirstHalf:", squiggleres4, squiggleres4[
5] - squiggleres4[3]
result = "Sequence"
else:
result = "Skip"
#print "This read don't match"
#print "Subset:",squiggleres2
#print "Subset Match Length:", squiggleres2[5]-squiggleres2[3]
#result = go_or_no(squiggleres[0],squiggleres[2],squiggleres[3],seqlen)
#print "result:",result
return result, channel_id, data.read_id, data.events[
0].start, squiggleres
except Exception, err:
err_string = "Time Warping Stuff : %s" % (err)
print >> sys.stderr, err_string
def mp_worker((channel_id, data, kmerhash, seqlen, readstarttime, kmerhash_subset)):
if (time.time() - readstarttime) > args.time:
print "We have a timeout"
return "timeout", channel_id, data.read_id, data.events[0].start
else:
try:
print "Read start time", readstarttime
squiggle = extractsquig(data.events)
squiggleres = squiggle_search2(squiggle, channel_id, data.read_id, kmerhash, seqlen)
print "Full Length:", squiggleres
print "Full Length Match Length:", squiggleres[5] - squiggleres[3]
squiggleres2 = squiggle_search2(squiggle, channel_id, data.read_id, kmerhash_subset, seqlen)
squiggleres3 = squiggle_search2(squiggle[100:200], channel_id, data.read_id, kmerhash_subset, seqlen)
squiggleres4 = squiggle_search2(squiggle[50:100], channel_id, data.read_id, kmerhash_subset, seqlen)
if (
squiggleres2[5] > squiggleres3[3] > squiggleres2[3]
and squiggleres3[3] > squiggleres4[3] > squiggleres2[3]
):
print "!!!!!!!!!!!!!!!! We got a good one! !!!!!!!!!!!!!!!!"
print "Subset:", squiggleres2
print "Subset Match Length:", squiggleres2[5] - squiggleres2[3]
print "SecondHalf:", squiggleres3
print "SecondHalf:", squiggleres3[5] - squiggleres3[3]
print "FirstHalf:", squiggleres4, squiggleres4[5] - squiggleres4[3]
result = "Sequence"
else:
result = "Skip"
# print "This read don't match"
# print "Subset:",squiggleres2
# print "Subset Match Length:", squiggleres2[5]-squiggleres2[3]
# result = go_or_no(squiggleres[0],squiggleres[2],squiggleres[3],seqlen)
# print "result:",result
return result, channel_id, data.read_id, data.events[0].start, squiggleres
except Exception, err:
err_string = "Time Warping Stuff : %s" % (err)
print >> sys.stderr, err_string
def process_hdf5((filename, seqids, threedarray, procampres, seqlen, args)):
returndict = dict()
hdf = h5py.File(filename, "r")
for read in hdf["Analyses"]["EventDetection_000"]["Reads"]:
events = hdf["Analyses"]["EventDetection_000"]["Reads"][read]["Events"][()]
event_collection = list()
for event in events:
event_collection.append(float(event[0]))
squiggle = event_collection[50:300]
# print seqlen
# squiggle,channel_id,read_id,args,seqids,threedarray,seqlen
squiggleres = squiggle_search2(squiggle, 0, 0, args, seqids, threedarray, seqlen)
if 1:
print squiggleres[0], squiggleres[2], squiggleres[3]
try:
# seqid,direction,position,seqlen,args,distance,refmatchlen,querymatchlen
# print squiggleres
# result = go_or_no2(squiggleres[0],squiggleres[2],squiggleres[3],seqlen,args,squiggleres[1],0,0)
result = go_or_no(squiggleres[0], squiggleres[2], squiggleres[3], seqlen, args)
except Exception, err:
print "ERROR"
print err
