def write_new_pairs(pair_file_path,new_pairs,qbed_file_new,sbed_path_new):
""" appends the new pairs to the end of the pair f:/wile and then changes it
moves it into qa file fmt"""
write_file = open(pair_file_path,'a')
for pair in new_pairs:
new_line = "{0}\t{1}\n".format(pair['qaccn'],pair['saccn'])
write_file.write(new_line)
write_file.close()
header = pair_file_path.split(".")[0]
raw_file = "{0}.raw.with_new.filtered".format(header)
pairs_to_qa(pair_file_path,'pair',qbed_file_new,sbed_path_new,raw_file)
def write_new_pairs(pair_file_path, new_pairs, qbed_file_new, sbed_path_new):
""" appends the new pairs to the end of the pair f:/wile and then changes it
moves it into qa file fmt"""
write_file = open(pair_file_path, 'a')
for pair in new_pairs:
new_line = "{0}\t{1}\n".format(pair['qaccn'], pair['saccn'])
write_file.write(new_line)
write_file.close()
header = pair_file_path.split(".")[0]
raw_file = "{0}.raw.with_new.filtered".format(header)
pairs_to_qa(pair_file_path, 'pair', qbed_file_new, sbed_path_new, raw_file)
def main(cns_file,
qdups_path,
sdups_path,
pair_file,
fmt,
qbed,
sbed,
qpad,
spad,
blast_path,
unmasked_fasta,
mask='F',
ncpu=8):
pool = Pool(ncpu)
bl2seq = "%s " % blast_path + \
"-p blastn -D 1 -E 2 -q -2 -r 1 -G 5 -W 7 -F %s " % mask + \
" -e %(e_value).2f -i %(qfasta)s -j %(sfasta)s \
-I %(qstart)d,%(qstop)d -J %(sstart)d,%(sstop)d | grep -v '#' \
| grep -v 'WARNING' | grep -v 'ERROR' "
qfastas = get_masked_fastas(qbed)
sfastas = get_masked_fastas(
sbed) if qbed.filename != sbed.filename else qfastas
################# file paths #####################
qnolocaldups_path = qbed.path.split(".")[0] + ".nolocaldups.bed"
snolocaldups_path = sbed.path.split(".")[0] + ".nolocaldups.bed"
qlocaldups_path = qbed.path.split(".")[0] + ".localdups"
slocaldups_path = sbed.path.split(".")[0] + ".localdups"
npair_file, nqlocaldups, nslocaldups, ncns_file = map(
make_copy_of_file,
[pair_file, qlocaldups_path, slocaldups_path, cns_file])
##########################################
qdups = parse_dups(qdups_path)
sdups = parse_dups(sdups_path)
dups, rdups = get_pairs(pair_file, fmt, qdups, sdups)
print len(dups), len(rdups)
ldups = get_large_dups(dups, qdups, sdups)
rdups_dic = defaultdict(dict)
rdups_both = [(qparent, sparent) for qparent, sparent in dups
if qparent in rdups and sparent in rdups]
for (qparent, sparent) in dups:
if skip_pair(qparent, sparent, rdups, rdups_both, ldups): continue
cnss_size = []
qfeat_dups = get_all_dups(qdups, qparent)
sfeat_dups = get_all_dups(sdups, sparent)
pairs = [True]
_get_dups_gen = get_dups(qfeat_dups, sfeat_dups, qbed, sbed)
def get_dups_gen():
try:
return _get_dups_gen.next()
except StopIteration:
return None
while any(pairs):
cnss_dups = []
pairs = [get_dups_gen() for i in range(ncpu)]
###this is for parellization#########
spad_map = [spad] * len(pairs)
qpad_map = [qpad] * len(pairs)
sfastas_map = [sfastas] * len(pairs)
qfastas_map = [qfastas] * len(pairs)
bl2seq_map = [bl2seq] * len(pairs)
###################################
cmds = [
c for c in map(get_cmd, [l for l in pairs if l], bl2seq_map,
qfastas_map, sfastas_map, qpad_map, spad_map)
if c
]
results = (
r for r in pool.map(commands.getoutput, [c[0] for c in cmds]))
for res, (cmd, qfeat, sfeat) in zip(results, cmds):
orient = qfeat['strand'] == sfeat['strand'] and 1 or -1
if not res.strip(): cnss = []
else:
cnss = parse_blast(res, orient, qfeat, sfeat, qbed, sbed,
qpad, spad, unmasked_fasta)
print >> sys.stderr, "(%i)" % len(cnss)
cnss_fmt = ",".join(map(lambda l: ",".join(map(str, l)), cnss))
cnss_size.append(
(len(cnss) * -1, qfeat["start"], sfeat["start"],
qfeat["accn"], sfeat["accn"], cnss_fmt))
pairs = [pairs[-1]]
######################################################################
if qparent in rdups:
if (qparent, sparent) in rdups_dic[qparent].keys():
logging.info((qparent, sparent))
rdups_dic[qparent].update({(qparent, sparent): cnss_size})
elif sparent in rdups:
if (qparent, sparent) in rdups_dic[sparent].keys():
logging.info((qparent, sparent))
rdups_dic[sparent].update({(qparent, sparent): cnss_size})
else:
cnss_size.sort()
cns_number, qfeat_start, sfeat_start, qaccn, saccn, largest_cnss = cnss_size[
0]
qfeat = qbed.accn(qaccn)
sfeat = sbed.accn(saccn)
print >> sys.stderr, "FINAL: {0},{1},{2}".format(
qaccn, saccn, cns_number)
write_new_dups(npair_file, ncns_file, nqlocaldups, nslocaldups,
cnss_size, qparent, sparent, qfeat, sfeat, qdups,
sdups)
best_reps = best_repeats(rdups_dic)
for dparents in best_reps.keys():
qparent, sparent = dparents
### one or list? cnss[0]?
cns_number, qfeat_start, sfeat_start, qaccn, saccn, largest_cnss = best_reps[
dparents]
qfeat = qbed.accn(qaccn)
sfeat = sbed.accn(saccn)
write_new_dups(npair_file, ncns_file, nqlocaldups, nslocaldups,
[best_reps[dparents]], qparent, sparent, qfeat, sfeat,
qdups, sdups)
write_nolocaldups(
qbed.path, nqlocaldups,
"{0}.nolocaldups.bed.local".format(qbed.path.split(".")[0]))
write_nolocaldups(
sbed.path, nslocaldups,
"{0}.nolocaldups.bed.local".format(sbed.path.split(".")[0]))
pairs_to_qa(npair_file, 'pair',
"{0}.nolocaldups.bed.local".format(qbed.path.split(".")[0]),
"{0}.nolocaldups.bed.local".format(sbed.path.split(".")[0]),
"{0}.raw.filtered.local".format(options.pairs.split(".")[0]))
def main(cns_file,qdups_path,sdups_path,pair_file,fmt,qbed,sbed,qpad,spad,blast_path,mask='F',ncpu=8):
pool = Pool(ncpu)
bl2seq = "%s " % blast_path + \
"-p blastn -D 1 -E 2 -q -2 -r 1 -G 5 -W 7 -F %s " % mask + \
" -e %(e_value).2f -i %(qfasta)s -j %(sfasta)s \
-I %(qstart)d,%(qstop)d -J %(sstart)d,%(sstop)d | grep -v '#' \
| grep -v 'WARNING' | grep -v 'ERROR' "
qfastas = get_masked_fastas(qbed)
sfastas = get_masked_fastas(sbed) if qbed.filename != sbed.filename else qfastas
################# file paths #####################
qnolocaldups_path = qbed.path.split(".")[0] + ".all.nolocaldups.bed"
snolocaldups_path = sbed.path.split(".")[0] + ".all.nolocaldups.bed"
qlocaldups_path = qbed.path.split(".")[0] + ".all.localdups"
slocaldups_path = sbed.path.split(".")[0] + ".all.localdups"
npair_file,nqlocaldups,nslocaldups, ncns_file = map(make_copy_of_file,[pair_file,qlocaldups_path,slocaldups_path,cns_file])
##########################################
qdups = parse_dups(qdups_path)
sdups = parse_dups(sdups_path)
dups,rdups = get_pairs(pair_file,fmt,qdups,sdups)
print len(dups), len(rdups)
ldups = get_large_dups(dups,qdups,sdups)
rdups_dic = defaultdict(dict)
rdups_both = [(qparent,sparent) for qparent,sparent in dups if qparent in rdups and sparent in rdups]
for (qparent,sparent) in dups:
if skip_pair(qparent,sparent,rdups,rdups_both,ldups):continue
cnss_size = []
qfeat_dups = get_all_dups(qdups,qparent)
sfeat_dups = get_all_dups(sdups,sparent)
pairs = [True]
_get_dups_gen = get_dups(qfeat_dups,sfeat_dups,qbed,sbed)
def get_dups_gen():
try: return _get_dups_gen.next()
except StopIteration: return None
while any(pairs):
cnss_dups = []
pairs = [get_dups_gen() for i in range(ncpu)]
###this is for parellization#########
spad_map = [spad] * len(pairs)
qpad_map = [qpad] * len(pairs)
sfastas_map = [sfastas] * len(pairs)
qfastas_map = [qfastas] * len(pairs)
bl2seq_map = [bl2seq] * len(pairs)
###################################
cmds = [c for c in map(get_cmd, [l for l in pairs if l],
bl2seq_map,qfastas_map,sfastas_map,qpad_map,spad_map) if c]
results = (r for r in pool.map(commands.getoutput, [c[0] for c in cmds]))
for res, (cmd, qfeat, sfeat) in zip(results, cmds):
orient = qfeat['strand'] == sfeat['strand'] and 1 or -1
if not res.strip(): cnss = []
else: cnss = parse_blast(res, orient, qfeat, sfeat, qbed, sbed, qpad,spad)
print >>sys.stderr, "(%i)" % len(cnss)
cnss_fmt = ",".join(map(lambda l: ",".join(map(str,l)),cnss))
cnss_size.append((len(cnss)*-1,qfeat["start"],sfeat["start"],qfeat["accn"],sfeat["accn"],cnss_fmt))
pairs = [pairs[-1]]
######################################################################
if qparent in rdups:
if (qparent,sparent) in rdups_dic[qparent].keys(): logging.info((qparent,sparent))
rdups_dic[qparent].update({(qparent,sparent):cnss_size})
elif sparent in rdups:
if (qparent,sparent) in rdups_dic[sparent].keys(): logging.info((qparent,sparent))
rdups_dic[sparent].update({(qparent,sparent):cnss_size})
else:
cnss_size.sort()
cns_number,qfeat_start,sfeat_start,qaccn,saccn,largest_cnss = cnss_size[0]
qfeat = qbed.accn(qaccn)
sfeat = sbed.accn(saccn)
print >>sys.stderr, "FINAL: {0},{1},{2}".format(qaccn,saccn,cns_number)
write_new_dups(npair_file,ncns_file,nqlocaldups,nslocaldups,cnss_size,qparent,sparent,qfeat,sfeat,qdups,sdups)
best_reps = best_repeats(rdups_dic)
for dparents in best_reps.keys():
#print dparents
qparent,sparent = dparents
#print parents,best_reps[parents]
### one or list? cnss[0]?
cns_number,qfeat_start, sfeat_start,qaccn,saccn,largest_cnss = best_reps[dparents]
qfeat= qbed.accn(qaccn)
sfeat = sbed.accn(saccn)
write_new_dups(npair_file,ncns_file,nqlocaldups,nslocaldups,[best_reps[dparents]],qparent,sparent,qfeat,sfeat,qdups,sdups)
write_nolocaldups(qbed.path,nqlocaldups,"{0}.all.nolocaldups.bed.local".format(qbed.path.split(".")[0]))
write_nolocaldups(sbed.path,nslocaldups,"{0}.all.nolocaldups.bed.local".format(sbed.path.split(".")[0]))
pairs_to_qa(npair_file,'pair',"{0}.all.nolocaldups.bed.local".format(qbed.path.split(".")[0]),"{0}.all.nolocaldups.bed.local".format(sbed.path.split(".")[0]),"{0}.raw.filtered.local".format(options.pairs.split(".")[0]))