def oldtails():
aligns = M5File(primary)
#where I'm putting the good hits
mapOut = open(outname, "w")
#where I'm putting the tails
tfq = NamedTemporaryFile(prefix="tails_", suffix=".fastq", delete=False, dir=basedir)
ALLTEMPFILES.append( tfq.name )
whichEnd = defaultdict(list)
#extract the tails
ntails = 0
for a in aligns:
if a.qstart >= MINTAIL:
tseq1 = reads[a.qname].subSeq(None, a.qstart)
#prolog
tseq1.name = "%s_::_5_::_%d,%d" % (tseq1.name, a.qstart, a.qseqlength)
tfq.write(str(tseq1))
ntails += 1
if a.qend - a.qseqlength > MINTAIL:
tseq2 = reads[a.qname].subSeq(a.qend, None)
#epilog
tseq2.name = "%s_::_3_::_%d,%d" % (tseq2.name, a.qend, a.qseqlength)
tfq.write(str(tseq2))
ntails += 1
mapOut.write(str(a)+"\n")
#don't want redundant hits on a single flank
whichEnd[a.qname].append(a.tname)
tfq.close()
logging.info("%d unmapped tails" % (ntails))
#map tails
tailAlign = NamedTemporaryFile(prefix="tails_", suffix=".m5", delete=False, dir=basedir)
tailAlign = tailAlign.name
ALLTEMPFILES.append(tailAlign)
blasr(tfq.name, target, nproc=nproc, bestn=1, outname=tailAlign)
aligns2 = M5File(tailAlign)
logging.info("%d tails mapped" % len(aligns2))
for a in aligns2:
#get the carryon info
name, direct, se = a.qname.split("_::_")
pos, length = map(int, se.split(','))
#correct it's information
a.qname = name
a.qseqlength = length
#prevent redundant flank map
if a.tname in whichEnd[a.qname]:
logging.info("%s failed ref map" % a.tname)
continue
whichEnd[a.qname].append(a.tname)
#epilogs need to be updated
if direct == '3':
a.qstart += pos
a.qend += pos
mapOut.write(str(a)+"\n")
mapOut.close()
return
def m5ToOvlGraph(readNames, fileName):
"""
Create the graph
"""
connector = AlignmentConnector()
alignments = M5File(fileName)
graph = nx.Graph()
#filt = []
#get only the single best alignment between any two reads
fdict = {}
for align in alignments:
if align.qname == align.tname:
continue
name = [align.qname, align.tname]
name.sort()
name = ":".join(name)
if name in fdict:
if align.score < fdict[name].score:
fdict[name] = align
else:
fdict[name] = align
alignments = fdict.values()
#make edges for all overlaps
for align in alignments:
if align.qname == align.tname:
continue
extend = connector.extendsTarget(align)
align.support = extend
if extend != SUPPORTFLAGS.none:
graph.add_edge(align.qname, align.tname, data=align)
return graph
def run(argv):
print argv
args = parseArgs(argv)
if args.m4.endswith("m5"):
aligns = M5File(args.m4)
else:
aligns = M4File(args.m4)
if args.reads.endswith("fasta"):
reads = FastaFile(args.reads)
elif args.reads.endswith("fastq"):
temp = FastqFile(args.reads)
reads = {}
for i in temp:
reads[i] = temp[i].seq
del (temp)
else:
logging.error("Expected Fasta or Fastq for READS (%s)" % args.reads)
exit(1)
logging.info("Extracting tails")
tailfastq = tempfile.NamedTemporaryFile(suffix=".fasta",
delete=False,
dir=args.temp)
tailfastq.close()
tailfastq = tailfastq.name
logging.debug("Tail read tmp file %s " % (tailfastq))
r, t, m = extractTails(aligns,
reads,
outFq=tailfastq,
minLength=args.minTail)
logging.info("Parsed %d reads" % (r))
logging.info("Found %d tails" % (t))
logging.info("%d reads had double tails" % (m))
if t == 0:
logging.info("No tails -- Exiting")
exit(0)
logging.info("Mapping Tails")
tailmap = tempfile.NamedTemporaryFile(suffix=".m4",
delete=False,
dir=args.temp)
tailmap.close()
tailmap = tailmap.name
logging.debug("Read map tmp file %s " % (tailmap))
mapTails(tailfastq,
args.ref,
nproc=args.nproc,
out=tailmap,
useSa=args.noSa)
logging.info("Consolidating alignments")
logging.debug("Final file %s " % (args.output))
n = uniteTails(aligns, tailmap, args.output, args.inplace)
logging.info("%d tails mapped" % (n))
def singleOverlapAssembly(alldata, args):
"""
"""
global ALLTEMPFILES
data = alldata.stats
reads = NamedTemporaryFile(prefix="sol_", suffix=".fasta", delete=False, dir=args.tempDir)
ALLTEMPFILES.append(reads.name)
e1Seq = data["extendSeq1"]; e2Seq = data["extendSeq2"]
reads.write(">%s\n%s\n>%s\n%s\n" % ("seq1", e1Seq, "seq2", e2Seq))
reads.close()
alignFn = NamedTemporaryFile(prefix="sol_",suffix=".m5", delete=False, dir=args.tempDir)
ALLTEMPFILES.append(alignFn.name)
blasr(reads.name, reads.name, nproc=args.nproc, outname=alignFn.name)
aligns = M5File(alignFn)
# find best hit between the two
connector = AlignmentConnector()
bestS = None
bestA = 0
for i in aligns:
if i.qname != i.tname:
if connector.extendsTarget(i):
if i.score < bestS:
bestA = i
bestS = i.score
if bestS is None:
logging.info("no overlap between extenders")
return
#any of these steps could fail --
#Ensure the hit is valid
#(if + + and sameStrand we are okay, if - + and not sameStrand we are okay)
if data["sameStrand"] == (bestA.tstrand == '0'):
logging.info("bad overlap between extenders")
return
con = consensus([bestA])
bestA = bestA[0]
#strand correction...
if bestA.qname == "seq1":
if bestA.tstrand == '1':
e2Seq = e2Seq[:bestA.tstart].translate(revComp)[::-1]
seq = e1Seq[:bestA.qstart] + con.sequence.translate(revComp)[::-1] + e2Seq
else:
seq = e1Seq[:bestA.qstart] + con.sequence + e2Seq[bestA.tend:]
else:
if bestA.tstrand == '1':
e2Seq = e2Seq[:bestA.qstart].translate(revComp)[::-1]
seq = e1Seq[:bestA.tstart] + con.sequence + e2Seq
else:
seq = e1Seq[:bestA.qstart] + con.sequence + e2Seq[bestA.tstart:]
return seq
def consensusCalling(self, spot, bam, reference, args):
"""
Make a consensus of all the reads in the region and identify all of the SVs in the region
"""
#
MAXNUMREADS = 100 #I don't think we'll need more than this many reads
MAXATTEMPTS = MAXNUMREADS / 2 #I don't feel like trying 100 times
SPANBUFFER = 100 #number of bases I want a read to span
chrom, start, end = spot.chrom, spot.start, spot.end
buffer = args.buffer
supportReads = []
spanReads = []
#Fetch reads and trim
totCnt = 0
for read in bam.fetch(chrom, max(0, start - buffer - SPANBUFFER),
end + buffer + SPANBUFFER):
if read.qname not in spot.varReads:
continue
seq, qual = self.readTrim(read, start - buffer, end + buffer)
if read.pos < start - SPANBUFFER and read.aend > end + SPANBUFFER:
spanReads.append((len(seq), seq, qual))
else:
supportReads.append((seq, qual))
totCnt += 1
if len(spanReads) == 0:
logging.debug("noone spans - consensus aborted. %s" % (str(spot)))
spot.tags["noSpan"] = True
return [spot]
spanReads.sort(reverse=True)
if len(spanReads) > MAXNUMREADS:
origSupportReads = [(x[1], x[2]) for x in spanReads[:MAXNUMREADS]]
elif len(spanReads) + len(supportReads) > MAXNUMREADS:
origSupportReads = [(x[1], x[2]) for x in spanReads
] + supportReads[:MAXNUMREADS - len(spanReads)]
else:
origSupportReads = [(x[1], x[2]) for x in spanReads] + supportReads
logging.debug("Alt reads: %d total, %d extra support" %
(totCnt, len(origSupportReads)))
mySpots = []
refReadId = 0
haveVar = False
#Attempt each spanRead until we get one that passes
#while refReadId < len(spanReads) and not haveVar and refReadId < MAXATTEMPTS:
#refread = spanReads[refReadId]
#supportReads = origSupportReads[:refReadId] + origSupportReads[refReadId+1:]
refReadId += 1
#read that spans most of the region goes first
#use the rest for cleaning
#building consensus sequence
foutreads = NamedTemporaryFile(suffix=".fasta")
qoutreads = open(foutreads.name + '.qual', 'w')
for id, i in enumerate(origSupportReads):
foutreads.write(">%d\n%s\n" % (id, i[0]))
qoutreads.write(">%d\n%s\n" %
(id, " ".join(str(ord(j) - 33) for j in i[1])))
foutreads.flush()
qoutreads.flush()
#foutref = NamedTemporaryFile(suffix=".fasta")
#foutref.write(">%s:%d-%d\n%s" % (spot.chrom, start, end, refread[1]))
#foutref.flush()
logging.debug("Making the contig....")
#run it through phrap
#make out.fasta and out.fasta.qual
#run phrap
#if asm -- consensus only
r, o, e = exe("phrap %s -minmatch 6 -minscore 20" % (foutreads.name),
timeout=3)
if r != 0: #failed
logging.warning('phrap failed ' + self.name)
logging.warning(o)
logging.warning(e)
return [
] #here is where I'd like to add just the no-consensus spot
results = mergeFastaQual(foutreads.name + ".contigs",
foutreads.name + ".contigs.qual")
if len(results) == 0:
logging.warning('no asm made ' + self.name)
return [
] #here is where I'd like to add just the no-consensus spot
logging.info('%d contigs made %s' % (len(results), self.name))
#then run it through consensus
logging.debug("Polishing contigs")
alignOut = NamedTemporaryFile(suffix=".m5")
blasr(foutreads.name,
foutreads.name + ".contigs",
format="-m 5",
nproc=1,
outname=alignOut.name)
# elif no asm and consensus only (faster)
if args.polish == "pbbanana":
aligns = M5File(alignOut.name)
con = ">con\n%s\n" % consensus(aligns).sequence
conName = "pbbanana"
elif args.polish == "pbdagcon":
logging.debug("pbdagcon is running")
#using minerrreads - 1 because one f them is already being used as seed!
r, con, e = exe("pbdagcon -c %d -t 0 %s" %
(max(0, args.minErrReads - 1), alignOut.name),
timeout=1)
#r, con, e = exe("pbdagcon %s" % (alignOut.name), timeout=2)
logging.debug("back from pbdagcon")
logging.debug((r, e))
#raw_input("press ent")
if con is not None:
con = con[con.index("\n") + 1:]
else:
con = ""
conName = "pbdagcon"
alignOut.close()
#foutref.close()
foutreads.close()
#we don't have a consensus - retry
if len(con) == 0:
logging.debug("Trying another seed read for consensus")
con = results.values()[0].seq
logging.debug("%s %d bp seq" % (conName, len(con.split('\n')[1])))
#try improving consensus
conOut = NamedTemporaryFile(suffix=".fasta")
conOut.write(con)
#conOut.close()
conOut.flush()
refOut = NamedTemporaryFile(suffix=".fasta")
#j = reference.fetch(chrom, max(0, start-buffer), end+buffer)
#fout = open("f****e.ref.fasta",'w')
#fout.write(j)
#fout.close()
refOut.write(">%s:%d-%d\n%s\n" % (chrom, start, end, \
reference.fetch(chrom, max(0, start-buffer), end+buffer)))
refOut.flush()
#map consensus to refregion
varSam = NamedTemporaryFile(suffix=".sam")
blasr(conOut.name, refOut.name, format="--sam", outname=varSam.name)
#consensus=False) -- would this help?
#or what if I fed it through leftalign?
sam = pysam.Samfile(varSam.name)
matches = 0.0
bases = 0.0
nReads = 0
mySpots = []
for read in sam:
nReads += 1
spot.tags["consensusCreated"] = True
for svstart, svsize, svtype, altseq in expandCigar(
read, args.minIndelSize, CONFIRMCOLLAPSE, True):
newspot = copy.deepcopy(spot)
if spot.svtype == svtype and svtype == "INS":
haveVar = True
newspot.start = svstart + start - buffer
newspot.end = svstart + start - buffer
newspot.tags["seq"] = altseq
newspot.size = svsize
gt, gq = genotype(newspot)
newspot.tags["GT"] = gt
newspot.tags["GQ"] = gq
mySpots.append(newspot)
elif spot.svtype == svtype and svtype == "DEL":
haveVar = True
newspot.start = svstart + start - buffer
newspot.end = svstart + svsize + start - buffer
newspot.size = -svsize
gt, gq = genotype(newspot)
newspot.tags["GT"] = gt
newspot.tags["GQ"] = gq
newspot.tags["seq"] = reference.fetch(
chrom, newspot.start, newspot.end)
mySpots.append(newspot)
#identity = matches/bases
#If no var, nothing is returned.
#for newspot in mySpots:
#newspot.tags["alnIdentityEstimate"] = identity
#Keep reporting the actual contigs out until we
#find a reason to need it (and also we can get quals...)
#vbam.reset()
#for id, read in enumerate(vbam):
#newspot.tags["contigSeq%d" % (id)] = read.seq
#newspot.tags["contigQual%d" % (id)] = read.qual
#vbam.close()
#varBam.close()
refOut.close()
logging.debug("%d consensus reads created %d spots" %
(nReads, len(mySpots)))
return mySpots
#!/usr/bin/python
import sys
from pbsuite.utils.FileHandlers import M4File, M5File
if __name__ == '__main__':
try:
fn = sys.argv[1]
except:
sys.stderr.write(("Error! Expected One Argument, " \
"an m4 or m5 alignment file\n"))
exit(1)
if fn.endswith('.m4'):
file = M4File(sys.argv[1])
elif fn.endswith('.m5'):
file = M5File(sys.argv[1])
else:
print "Unrecognized File Type (expecting .m4 or .m5)"
exit(1)
if len(sys.argv) == 3:
out = open(sys.argv[2], 'w')
else:
out = sys.stdout
out.write("\n".join(map(lambda x: x.toBed(), file)) + "\n")
if __name__ == '__main__':
args = parseArgs()
alignFile = args.outname+".m5"
consensusFile = args.outname+".fasta"
#extract the read I'm looking for
if args.target is not None:#Name
tempOut = open("temp.fasta",'w')
fasta = FastaFile(args.reads)
tempOut.write(">%s\n%s\n" % (args.target, fasta[args.target]))
tempOut.write
blasr(args.reads, tempOut.name, nproc=args.nproc, outName=alignFile)
aligns = M5File(alignFile)
fout = open(consensusFile, 'w')
results = consensus(aligns)
fout.write(">pbjpolish_%d_vote_%d_len\n" % (results.contribBases,\
results.fillBases, results.sequence))
#fout.write(">\n%s\n" % consensus(aligns))
fout.close()
elif args.Target is not None:#File
blasr(args.reads, args.Target, nproc=args.nproc, outName=alignFile)
aligns = M5File(alignFile)
fout = open(consensusFile, 'w')
results = consensus(aligns)
fout.write(">pbjpolish_%d_vote_%d_len\n%s\n" % (results.contribBases,\
results.fillBases, results.sequence))
def preunitereads(inputFastq, args):
"""
sent query, I'm going to pop all of the united reads onto this
"""
global ALLTEMPFILES
alignFile = NamedTemporaryFile(prefix="uni_",
suffix=".m5",
delete=False,
dir=args.tempDir).name
ALLTEMPFILES.append(alignFile)
readFile = NamedTemporaryFile(prefix="uni_",
suffix=".fasta",
delete=False,
dir=args.tempDir)
ALLTEMPFILES.append(readFile.name)
input = FastqFile(inputFastq)
for read in input:
readFile.write(">%s\n%s\n" % (input[read].name, input[read].seq))
readFile.close()
readFile = readFile.name
blasr(readFile,
readFile,
bestn=5,
nCandidates=20,
nproc=args.nproc,
outname=alignFile)
aligns = M5File(alignFile)
con = AlignmentConnector()
extenders = []
for a in aligns:
if a.tname == a.qname:
continue
if a.qstart - a.qend < 500 or a.tstart - a.tend < 500:
continue
sup = con.extendsTarget(a, minCovers=500)
#sup = con.extendsTarget(a, minCovers=100)
a.support = sup
if sup in [SUPPORTFLAGS.left, SUPPORTFLAGS.right]:
extenders.append(a)
best = {} #best of queries
for i in extenders:
score = 0
if i.qname in best:
score = best[i.qname].score
if i.score < score:
best[i.qname] = i
#print "q"
#for i in best.values():
#print str(i)
best2 = {} #best of targets
for i in best.values():
score = 0
if i.tname in best2:
score = best2[i.tname].score
if i.score < score:
best2[i.tname] = i
#print "t"
#for i in best2.values():
#print str(i)
best3 = {} #best of both
for i in best2.values():
keys = [i.qname, i.tname]
keys.sort()
keys = "".join(keys)
score = 0
if keys in best3:
score = best3[keys].score
if i.score < score:
best3[keys] = i
#print 'b'
#for i in best3.values():
#print str(i)
reads = FastqFile(inputFastq)
fout = open(inputFastq, 'a')
count = 0
for i in best3.values():
qseq = None
if i.support == SUPPORTFLAGS.left:
if i.qstrand == '0':
qseq = reads[i.qname].seq + reads[i.tname].seq[i.tend:]
elif i.qstrand == '1':
qseq = reads[i.qname].seq + reads[
i.tname].seq[i.tend:].translate(revComp)
if i.support == SUPPORTFLAGS.right:
if i.qstrand == '0':
qseq = reads[i.tname].seq[:i.tstart] + reads[i.qname].seq
elif i.qstrand == '1':
qseq = reads[i.tname].seq[:i.tstart].translate(
revComp) + reads[i.qname].seq
if qseq is not None:
count += 1
fout.write("@%s_%s\n%s\n+\n%s\n" %
(i.qname, i.tname, qseq, "!" * len(qseq)))
logging.info("Preunited %d reads" % (count))
fout.close()
def buildFillSeq(data, inputReads, args):
"""
Using all of the information in the namedtuple returned from getSubSeqs,
go through the process of building the filling sequence.
load the filling sequence in to the data
"""
#try to build span
if SUPPORTFLAGS.span in data.stats["support"][0]:
logging.debug("build span")
alignFile = NamedTemporaryFile(prefix="scon_",
suffix=".m5",
delete=False,
dir=args.tempDir)
alignFile.close()
alignFile = alignFile.name
ALLTEMPFILES.append(alignFile)
#blasr(data.spanReads, data.spanSeed, bestn = 1, nproc = args.nproc, outname=alignFile)
blasr(inputReads,
data.spanSeed,
bestn=1,
nproc=args.nproc,
outname=alignFile)
aligns = M5File(alignFile)
if len(aligns) > 0:
con = consensus(aligns)
#if successful we're done
if con.contribBases > 0 and con.fillBases > 0: #must be
sequence = con.sequence #strandCorrector(data.stats["spanSeedStrand1"], con.sequence)
data.stats["fillSeq"] = sequence
data.stats["contribSeqs"] = con.contribSeqs
data.stats["contribBases"] = con.contribBases
data.stats["fillBases"] = con.fillBases
return
else:
logging.info("no mapping... picking span seq")
sequence = FastaFile(data.spanSeed).values()[0]
data.stats["fillSeq"] = sequence
data.stats["contribSeqs"] = 1
data.stats["contribBases"] = len(sequence)
data.stats["fillBases"] = len(sequence)
return
#no span -- we need to do flanks
flank1Success = False
flank2Success = False
logging.debug(json.dumps(data.stats, indent=4))
fl1Flag = SUPPORTFLAGS.left if data.stats["seed1"].endswith(
"e5") else SUPPORTFLAGS.right
if data.stats["seed2"] is not None:
fl2Flag = SUPPORTFLAGS.left if data.stats["seed2"].endswith(
"e5") else SUPPORTFLAGS.right
else:
fl2Flag = None
logging.debug((fl1Flag, fl2Flag))
if fl1Flag in data.stats["support"][1]:
logging.debug("build flank1 %d" % fl1Flag)
alignFile = NamedTemporaryFile(prefix="f1con_",
suffix=".m5",
delete=False,
dir=args.tempDir)
alignFile.close()
alignFile = alignFile.name
ALLTEMPFILES.append(alignFile)
#blasr(data.flank1Reads, data.flank1Seed, bestn=1, nproc=args.nproc, outname=alignFile)
blasr(inputReads,
data.flank1Seed,
bestn=1,
nproc=args.nproc,
outname=alignFile)
aligns = M5File(alignFile)
if len(aligns) > 0:
con = consensus(aligns)
if con.contribBases > 0 and con.fillBases > 0: #must be
sequence = con.sequence #strandCorrector(data.stats["extendF1SeedStrand"], con.sequence)
data.stats["extendSeq1"] = sequence
data.stats["contribSeqs"] += con.contribSeqs
data.stats["contribBases"] += con.contribBases
data.stats["fillBases"] += con.fillBases
flank1Success = True
else:
logging.info("no mapping... picking f1 seq")
sequence = FastaFile(data.flank1Seed).values()[0]
data.stats["extendSeq1"] = sequence
data.stats["contribSeqs"] = 1
data.stats["contribBases"] = len(sequence)
data.stats["fillBases"] = len(sequence)
flank1Success = True
if fl2Flag in data.stats["support"][2]:
logging.debug("build flank2 %d" % fl2Flag)
alignFile = NamedTemporaryFile(prefix="f2con_",
suffix=".m5",
delete=False,
dir=args.tempDir)
alignFile.close()
alignFile = alignFile.name
ALLTEMPFILES.append(alignFile)
#blasr(data.flank2Reads, data.flank2Seed, bestn=1, nproc=args.nproc, outname=alignFile)
blasr(inputReads,
data.flank2Seed,
bestn=1,
nproc=args.nproc,
outname=alignFile)
aligns = M5File(alignFile)
if len(aligns) > 0:
con = consensus(aligns)
if con.contribBases > 0 and con.fillBases > 0: #must be
sequence = con.sequence #strandCorrector(data.stats["extendF2SeedStrand"], con.sequence)
data.stats["extendSeq2"] = sequence
data.stats["contribSeqs"] += con.contribSeqs
data.stats["contribBases"] += con.contribBases
data.stats["fillBases"] += con.fillBases
flank2Success = True
else:
logging.info("no mapping... picking f1 seq")
sequence = FastaFile(data.flank2Seed).values()[0]
data.stats["extendSeq2"] = sequence
data.stats["contribSeqs"] = 1
data.stats["contribBases"] = len(sequence)
data.stats["fillBases"] = len(sequence)
flank2Success = True
if flank1Success and flank2Success:
logging.debug("mid unite")
seq = singleOverlapAssembly(data, args)
if seq is not None:
data.stats["fillSeq"] = seq
return
import sys
from pbsuite.utils.FileHandlers import FastqFile, M5File
from pbsuite.jelly.Support import AlignmentConnector, SUPPORTFLAGS
"""
Need to do work here
"""
if __name__ == '__main__':
connector = AlignmentConnector()
aligns = connector.parseAlignments(M5File(sys.argv[1]))
reads = FastqFile(sys.argv[2])
bestScore = None
best = None
fout = open("reads.fastq",'w')
spanCount = 0
for readGroup in aligns:
if readGroup[0].qname.startswith("ref"):
continue
if len(readGroup) == 2:
r1, r2 = readGroup
a = connector.extendsTarget(r1)
b = connector.extendsTarget(r2)
if a != SUPPORTFLAGS.none and b != SUPPORTFLAGS.none:
spanCount += 1
print r1.qname, "spans"
rStart = min(r1.qend, r2.qend)
rEnd = max(r1.qstart, r2.qstart)
t = reads[r1.qname].subSeq(rStart, rEnd)
def consensusCalling(self, spot, bam, reference, args):
"""
Make a consensus of all the reads in the region and identify all of the SVs in the region
"""
#
MAXNUMREADS = 100 #I don't think we'll need more than this many reads
MAXATTEMPTS = 5 #MAXNUMREADS/2 #I don't feel like trying 100 times
SPANBUFFER = 100 #number of bases I want a read to span
chrom, start, end = spot.chrom, spot.start, spot.end
buffer = args.buffer
supportReads = []
spanReads = []
#Fetch reads and trim
totCnt = 0
for read in bam.fetch(chrom, max(0, start - buffer - SPANBUFFER),
end + buffer + SPANBUFFER):
if read.qname not in spot.varReads:
continue
seq, qual = self.readTrim(read, start - buffer, end + buffer)
if read.pos < start - SPANBUFFER and read.aend > end + SPANBUFFER:
sz = spot.varReadsSize[spot.varReads.index(read.qname)]
spanReads.append((abs(sz - spot.tags["szMedian"]), seq, qual))
else:
supportReads.append((seq, qual))
totCnt += 1
if len(spanReads) == 0:
logging.debug("noone spans - consensus aborted. %s" % (str(spot)))
spot.tags["noSpan"] = True
return [spot]
#spanReads.sort(reverse=True)
spanReads.sort()
if len(spanReads) > MAXNUMREADS:
origSupportReads = [(x[1], x[2]) for x in spanReads[:MAXNUMREADS]]
elif len(spanReads) + len(supportReads) > MAXNUMREADS:
origSupportReads = [(x[1], x[2]) for x in spanReads
] + supportReads[:MAXNUMREADS - len(spanReads)]
else:
origSupportReads = [(x[1], x[2]) for x in spanReads] + supportReads
mySpots = []
refReadId = 0
haveVar = False
#Attempt each spanRead until we get one that passes
while refReadId < len(
spanReads) and not haveVar and refReadId < MAXATTEMPTS:
refread = spanReads[refReadId]
supportReads = origSupportReads[:refReadId] + origSupportReads[
refReadId + 1:]
refReadId += 1
#read that spans most of the region goes first
#use the rest for cleaning
#building consensus sequence
foutreads = NamedTemporaryFile(suffix=".fastq")
for id, i in enumerate(supportReads):
foutreads.write("@%d\n%s\n+\n%s\n" % (id, i[0], i[1]))
foutreads.flush()
foutref = NamedTemporaryFile(suffix=".fasta")
foutref.write(">%s:%d-%d\n%s" %
(spot.chrom, start, end, refread[1]))
foutref.flush()
alignOut = NamedTemporaryFile(suffix=".m5")
logging.debug("making the contig....")
#run it through phrap
#then run it through consensus
blasr(foutreads.name,
foutref.name,
format="-m 5",
nproc=1,
outname=alignOut.name)
if args.consensus == "pbbanana":
aligns = M5File(alignOut.name)
con = ">con\n%s\n" % consensus(aligns).sequence
conName = "pbbanana"
elif args.consensus == "pbdagcon":
logging.debug("pbdagcon is running")
#using minerreads - 1 because one f them is already being used as seed!
#I want to be sure I get something out... so just require somebody on there
#r, con, e = exe("pbdagcon -c %d -t 0 %s" % (1, alignOut.name), timeout=1)
#r, con, e = exe("pbdagcon -m 100 -c %d -t 0 %s" % (max(args.minErrReads - 1, 0), alignOut.name), timeout=1)
r, con, e = exe("pbdagcon -m 100 -c %d -t 0 %s" %
(3, alignOut.name),
timeout=1)
logging.debug("back from pbdagcon")
logging.debug((r, e))
#raw_input("press ent")
if con is not None:
con = con[con.index("\n") + 1:]
else:
con = ""
conName = "pbdagcon"
alignOut.close()
foutref.close()
foutreads.close()
#we don't have a consensus - retry
if len(con) == 0:
logging.debug("Trying another seed read for consensus")
continue
logging.debug("%s %d bp seq" % (conName, len(con.split('\n')[1])))
#try improving consensus
conOut = NamedTemporaryFile(suffix=".fasta")
conOut.write(con)
#conOut.close()
conOut.flush()
refOut = NamedTemporaryFile(suffix=".fasta")
#j = reference.fetch(chrom, max(0, start-buffer), end+buffer)
#fout = open("f****e.ref.fasta",'w')
#fout.write(j)
#fout.close()
refOut.write(">%s:%d-%d\n%s\n" % (chrom, start, end, \
reference.fetch(chrom, max(0, start-(buffer*2)), end+(buffer*2))))
refOut.flush()
#map consensus to refregion
varSam = NamedTemporaryFile(suffix=".sam")
blasr(conOut.name, refOut.name, format="-sam", outname=varSam.name,\
consensus=False) #-- would this help?
#or what if I fed it through leftalign?
#os.system("cp %s ." % (refOut.name))
#os.system("cp %s ." % (varSam.name))
sam = pysam.Samfile(varSam.name)
matches = 0.0
bases = 0.0
nReads = 0
minVarDiff = 10000
for read in sam:
localSpots = []
nReads += 1
spot.tags["consensusCreated"] = True
for svstart, svsize, svtype, altseq in expandCigar(
read, args.minIndelSize, CONFIRMCOLLAPSE, True):
newspot = copy.deepcopy(spot)
if spot.svtype == svtype and svtype == "INS":
#haveVar = True
newspot.start = svstart + start - (buffer * 2)
newspot.end = svstart + start - (buffer * 2)
newspot.tags["seq"] = altseq
newspot.size = svsize
gt, gq = genotype(newspot)
newspot.tags["GT"] = gt
newspot.tags["GQ"] = gq
if abs(spot.tags["szMedian"] -
newspot.size) < minVarDiff:
minVarDiff = abs(spot.tags["szMedian"] -
newspot.size)
if args.reportContig:
newspot.tags["contigseq"] = read.seq
newspot.tags["contigqual"] = read.qual
localSpots.append(newspot)
elif spot.svtype == svtype and svtype == "DEL":
#haveVar = True
newspot.start = svstart + start - (buffer * 2)
newspot.end = svstart + svsize + start - (buffer * 2)
newspot.size = svsize
gt, gq = genotype(newspot)
newspot.tags["GT"] = gt
newspot.tags["GQ"] = gq
newspot.tags["seq"] = reference.fetch(
chrom, newspot.start, newspot.end)
if abs(spot.tags["szMedian"] -
newspot.size) < minVarDiff:
minVarDiff = abs(spot.tags["szMedian"] -
newspot.size)
if args.reportContig:
newspot.tags["contigseq"] = read.seq
newspot.tags["contigqual"] = read.qual
localSpots.append(newspot)
if len(localSpots) > 0:
mySpots.append((minVarDiff, localSpots))
#identity = matches/bases
#If no var, nothing is returned.
#for newspot in mySpots:
#newspot.tags["alnIdentityEstimate"] = identity
#Keep reporting the actual contigs out until we
#find a reason to need it (and also we can get quals...)
#vbam.reset()
#for id, read in enumerate(vbam):
#newspot.tags["contigSeq%d" % (id)] = read.seq
#newspot.tags["contigQual%d" % (id)] = read.qual
#vbam.close()
#varBam.close()
refOut.close()
#logging.debug("%d consensus reads created %d spots" % (nReads, len(localSpots)))
if len(mySpots) == 0:
return []
mySpots.sort()
return mySpots[0][1]