def run_infernal(cmfile,
rnd,
seqs,
outfolder,
cpus=1,
score=0.0,
calibrate=False):
if exists("%s/R%ihits.fna" % (outfolder, rnd)):
return
if not exists(cmfile):
raise IOError("cmfile path provided does not exist: %s" % cmfile)
params = {
'--mid': True,
'--Fmid': 0.0002,
'--notrunc': True,
'--toponly': True,
'--cpu': cpus
} # '-g': True,
if calibrate:
calibrate_file(cmfile, cpus=cpus)
result = cmsearch_from_file(cmfile, seqs, RNA, cutoff=score, params=params)
with open("%s/R%ihits.fna" % (outfolder, rnd), 'w') as fout:
for hit in result:
fout.write(">%s score:%0.1f e-val:%f\n%s\n" %
(hit[0], hit[14], hit[15], seqs.getSeq(hit[0])))
if exists("%s/log.txt" % outfolder):
with open("%s/log.txt" % outfolder, 'a') as fout:
fout.write("Round %i: %i hits\n" % (rnd, len(result)))
def run_infernal(lock, cmfile, rnd, basefolder, outfolder, cpus=1, score=0.0, mpi=False):
try:
seqs = 0
#Only search unique sequences to save time
#check if previous run has removed some sequences, load correct file
if exists(basefolder + "R" + str(rnd) + "/R" + str(rnd) + "-Unique-Remaining.fasta"):
seqs = LoadSeqs(basefolder + "R" + str(rnd) + "/R" + str(rnd) + "-Unique-Remaining.fasta", moltype=RNA, aligned=False)
else:
seqs = LoadSeqs(basefolder + "R" + str(rnd) + "/R" + str(rnd) + "-Unique.fasta", moltype=RNA, aligned=False)
params = {'--mid': True, '--Fmid': 0.0002, '--notrunc': True, '--toponly': True, '--cpu': cpus} # '-g': True,
if mpi:
params['mpi'] = True
result = cmsearch_from_file(cmfile, seqs, RNA, cutoff=score, params=params)
fout = open(outfolder + "/R" + str(rnd) + "hits.txt", 'w')
fout.write(str(len(result)) + " hits\nheader,bitscore,e-value\n")
for hit in result:
fout.write(hit[0] + "," + str(hit[14]) + "," + str(hit[15]) + "\n")
fout.close()
lock.acquire()
fout = open(outfolder + "/log.txt", 'a')
fout.write("Round " + str(rnd) + ": " + str(len(result)) + " hits\n")
fout.close()
lock.release()
except Exception, e:
print str(e)
lock.release()
alnout = open(otufolder + "/locarnap-aln.sto", "w")
struct_dict = {"SS_cons": struct}
alnout.write(stockholm_from_alignment(aln, GC_annotation=struct_dict))
alnout.close()
print struct
# CLUSTER THE SECONDA
print "Creating CM and running Infernal over all rounds"
# create the cm file. Could call cmsearch_from_alignment but dont want to build
# cm file multiple times since is time consuming and processor intensive
cmfile = cmbuild_from_alignment(aln, struct, calibrate=True)
for i in range(7, 0, -1):
# run cmsearch over every round of SELEX
# Only search unique sequences to save time
seqs = LoadSeqs(
"/Users/Ely/Desktop/Ely_selection/R" + str(i) + "/R" + str(i) + "-Unique.fasta",
moltype=RNA,
aligned=False,
)
score = log2(seqs.getNumSeqs() * len(seqs))
print "R" + str(i) + " (" + str(score) + "):"
args = {"--toponly": True}
result = cmsearch_from_file(cmfile.name, seqs, RNA, cutoff=score, params=args)
print str(len(result)) + " hits"
fout = open(otufolder + "/R" + str(i) + "hits.txt", "w")
fout.write("header,bitscore,e-value\n")
for hit in result:
fout.write(hit[0] + "," + str(hit[14]) + "," + str(hit[15]) + "\n")
fout.close()
# remove found sequences from the round files
# cm file multiple times since is time consuming and processor intensive
cmfile = cmbuild_from_alignment(aln, struct, calibrate=True)
for i in range(7, 0, -1):
# run cmsearch over every round of SELEX
# Only search unique sequences to save time
seqs = 0
# check if previous run has removed some sequences, load correct file
if exists(args.f + "R" + str(i) + "/R" + str(i) + "-Unique-Remaining.fasta"):
print "Previous round run detected, runnning over remaining seqs"
seqs = LoadSeqs(
args.f + "R" + str(i) + "/R" + str(i) + "-Unique-Remaining.fasta", moltype=RNA, aligned=False
)
else:
seqs = LoadSeqs(args.f + "R" + str(i) + "/R" + str(i) + "-Unique.fasta", moltype=RNA, aligned=False)
result = cmsearch_from_file(
cmfile.name, seqs, RNA, cutoff=infernalscore, params={"--toponly": True, "--cpu": args.c}
)
print "R" + str(i) + ": " + str(len(result)) + " hits"
fout = open(currotufolder + "/R" + str(i) + "hits.txt", "w")
fout.write("header,bitscore,e-value\n")
for hit in result:
fout.write(hit[0] + "," + str(hit[14]) + "," + str(hit[15]) + "\n")
fout.close()
# clean up by removing cm file
remove(cmfile.name)
# remove found sequences from the round files
print "Runtime: " + str(time() - secs) + "m"
currgroup += 1
# skip group if less than 100 sequences
else:
print "Group has less than 100 sequences, skipping infernal run"