if __name__ == "__main__":
multiprocessing.freeze_support()
p = multiprocessing.Pool(args.procs)
manager = multiprocessing.Manager()
amplicon_file = open(args.ids, "r")
amplicons = []
for line in amplicon_file.readlines():
amplicons.append(line.rstrip())
if (args.verbose is True):
print amplicons
amplicon_file.close()
fasta_file = args.fasta
model_file_template = args.temp_model
model_file_complement = args.comp_model
model_kmer_means_template,tempkmerlen=process_model_file(model_file_template)
model_kmer_means_complement,compkmerlen=process_model_file(model_file_complement)
kmerhashT = process_ref_fasta_raw(fasta_file,model_kmer_means_template,tempkmerlen)
kmerhashC = process_ref_fasta_raw(fasta_file,model_kmer_means_complement,compkmerlen)
seqlengths = get_seq_len(fasta_file)
get_amplicons()
ampdict=[]
ampstartdict=dict()
ampenddict=dict()
counter = 0
procampres=manager.dict()
for amplicon in amplicons:
counter+=1
ampstart = int(float(amplicon.split(':', 1 )[1].split('-',1)[0]))
if not os.path.isdir(args.watchdir):
logger.error("Folder" + args.watchdir + " cannot be found. Exiting.")
sys.exit()
# Multiprocess setup
p = multiprocessing.Pool(args.procs)
manager = multiprocessing.Manager()
procampres = manager.dict()
fasta_file = args.fasta
seqlen = get_seq_len(fasta_file)
# Process model and reference fasta
model_file = args.temp_model
global model_kmer_means
global kmer_len
model_kmer_means, kmer_len = process_model_file(model_file)
seqids, threedarray = process_ref_fasta(fasta_file, model_kmer_means,
kmer_len)
# Scrap filenames
d = list()
filenamecounter = 0
for filename in glob.glob(os.path.join(args.watchdir, '*.fast5')):
filenamecounter += 1
d.append([filename, seqids, threedarray, procampres, seqlen, args])
for filename in glob.glob(os.path.join(args.watchdir, "pass", '*.fast5')):
filenamecounter += 1
d.append([filename, seqids, threedarray, procampres, seqlen, args])
for filename in glob.glob(os.path.join(args.watchdir, "fail", '*.fast5')):
filenamecounter += 1
d.append([filename, seqids, threedarray, procampres, seqlen, args])
for record in SeqIO.parse(args.fasta, 'fasta'):
print record.id
idlist.append(record.id)
idsconcat = " ".join(idlist)
for amplicon in amplicons:
if amplicon.split(':')[0] not in idsconcat:
print "!** At least one amplicon is not in your reference sequence.\n\r Please check:"
print amplicon
print "!** This program will now exit.\n"
sys.exit()
fasta_file = args.fasta
model_file_template = args.temp_model
model_kmer_means_template, kmer_len = process_model_file(
model_file_template)
kmerhashT = process_ref_fasta_raw(fasta_file, model_kmer_means_template)
seqlengths = get_seq_len(fasta_file)
get_amplicons()
ampdict = []
ampstartdict = dict()
ampenddict = dict()
correctedampdict = []
correctedampstartdict = dict()
correctedampenddict = dict()
counter = 0
procampres = manager.dict()
print record.id
idlist.append(record.id)
idsconcat = " ".join(idlist)
for amplicon in amplicons:
if amplicon.split(':')[0] not in idsconcat:
print "!** At least one amplicon is not in your reference sequence.\n\r Please check:"
print amplicon
print "!** This program will now exit.\n"
sys.exit()
fasta_file = args.fasta
model_file_template = args.temp_model
model_kmer_means_template,kmer_len=process_model_file(model_file_template)
kmerhashT = process_ref_fasta_raw(fasta_file,model_kmer_means_template)
seqlengths = get_seq_len(fasta_file)
get_amplicons()
ampdict=[]
ampstartdict=dict()
ampenddict=dict()
correctedampdict=[]
correctedampstartdict=dict()
correctedampenddict=dict()
counter = 0
procampres=manager.dict()
if __name__ == "__main__":
multiprocessing.freeze_support()
p = multiprocessing.Pool(args.procs)
manager = multiprocessing.Manager()
amplicon_file = open(args.ids, "r")
amplicons = []
for line in amplicon_file.readlines():
amplicons.append(line.rstrip())
if (args.verbose is True):
print amplicons
amplicon_file.close()
fasta_file = args.fasta
model_file_template = args.temp_model
model_file_complement = args.comp_model
model_kmer_means_template, tempkmerlen = process_model_file(
model_file_template)
model_kmer_means_complement, compkmerlen = process_model_file(
model_file_complement)
kmerhashT = process_ref_fasta_raw(fasta_file, model_kmer_means_template,
tempkmerlen)
kmerhashC = process_ref_fasta_raw(fasta_file, model_kmer_means_complement,
compkmerlen)
seqlengths = get_seq_len(fasta_file)
get_amplicons()
ampdict = []
ampstartdict = dict()
ampenddict = dict()
counter = 0
procampres = manager.dict()
checkfasta(args.fasta)
if not os.path.isdir(args.watchdir):
print "**! Sorry, but the folder " + args.watchdir + " cannot be found.\n\n**! Please check you have entered the path correctly and try again.\n\n**! This script will now terminate.\n"
sys.exit()
p = multiprocessing.Pool(args.procs)
manager = multiprocessing.Manager()
procampres = manager.dict()
fasta_file = args.fasta
seqlen = get_seq_len(fasta_file)
model_file = args.temp_model
global model_kmer_means
global kmer_len
model_kmer_means, kmer_len = process_model_file(model_file)
seqids, threedarray = process_ref_fasta(fasta_file, model_kmer_means, kmer_len)
# print "init kmerhash",type(kmerhash)
# print type(threedarray)
# model_file = args.temp_model
# model_kmer_means,kmer_len=process_model_file(model_file)
# kmerhash_subset = process_ref_fasta_subset(fasta_file,model_kmer_means,seqlen,kmer_len)
# kmerhash_subset = process_ref_fasta(fasta_file,model_kmer_means,seqlen,kmer_len)
# sys.exit()
d = list()
filenamecounter = 0
