def plotBaseBias(self, nCores=1, verbose=False, pp=""):
"""
Plots the base bias, or per base sequence content.
Parameters
=============
nCores INT Number of cpu cores to use. -1 to use all cores. (Default: 1)
verbose BOOL Print runtime (Default: False)
pp PdfPage Only used by the runQC function.
"""
if not "maxLen" in self.__dict__.keys():
print "Running readLength"
self.readLength(printOut=False)
bArray = np.zeros((self.maxLen, 5), dtype=np.uint32)
if nCores == 1:
count = 0
start = time.clock()
for seq, qual in cFileGen(self.inFile):
tmpBases[range(len(seq)), map(lambda y: baseDict[y], seq)] += 1
count += 1
if not count % 100000:
print "Finished %d of %d reads" % (count, self.numReads)
if verbose:
print "CPU time: %.3f seconds" % (time.clock() - start)
else:
nCores = setCores(nCores, verbose)
p = [] # process array
pConns = [] # parent connection array
for i in xrange(nCores):
pConn, cConn = Pipe() # returns (parent connection, child connection)
pConns.append(pConn)
# initialize processes
p.append(Process(target=bbWorker, args=(self.inFile, self.maxLen, i, nCores, cConn)))
wallStart = time.time()
for i in xrange(nCores): # start processes
p[i].start()
cpuTotal = 0
for i in xrange(nCores):
tmpBases, cpuTime = pConns[i].recv() # get results from processes
bArray += tmpBases
cpuTotal += cpuTime
wallTime = time.time() - wallStart
for i in xrange(nCores):
p[i].join()
if verbose:
print "CPU time: %.3f seconds" % (cpuTotal)
print "Walltime: %.3f seconds" % (wallTime)
sums = np.array(np.sum(bArray, axis=1), dtype=np.float)
plt.figure(figsize=(12, 4))
for i in range(5):
plt.plot(bArray[:, i] / sums)
plt.legend(bases, loc=5, bbox_to_anchor=(1.1, 0.5))
plt.title("%s Base Bias" % (self.inFile.split("/")[-1]))
plt.ylabel("% of Bases")
plt.subplots_adjust(left=0.05, right=0.91)
if pp:
pp.savefig()
else:
plt.show()
def kmerWorker(inFile, k, wid, procs, cConn):
kmerDict = Counter()
cpuStart = time.clock()
count = 0
for seq, qual in cFileGen(inFile):
if count % procs == wid:
# tmp = [seq[i:i+k] for i in xrange(len(seq)-(k-1))]
tmp = [seq[i : i + k] for i in xrange(0, len(seq) - (k - 1), 3)]
for i in tmp:
if not "N" in i:
kmerDict[i] += 1
count += 1
cpuTime = time.clock()
cConn.send((kmerDict.most_common(100), cpuTime))
cConn.close()
def bbWorker(inFile, maxLen, wid, procs, cconn):
"""
base bias worker called by plotBaseBias for parallel computation
"""
tmpBases = np.zeros((maxLen, 5), dtype=np.uint32)
count = 0
cpuStart = time.clock()
# for seq, qual in fileGen(inFile):
for seq, qual in cFileGen(inFile):
if count % procs == wid:
tmpBases[range(len(seq)), map(lambda y: baseDict[y], seq)] += 1
count += 1
cpuTime = time.clock() - cpuStart
cconn.send((tmpBases, cpuTime))
cconn.close()
def qualWorker(inFile, maxLen, wid, procs, cConn):
"""
Quality worker called by plotQuality
"""
count = 0
quals = initMatrix(maxLen)
cpuStart = time.clock()
# for seq,qual in fileGen(inFile):
for seq, qual in cFileGen(inFile):
if count % procs == wid:
tmp = map(ord, qual)
for i in xrange(len(tmp)):
quals[i].append(tmp[i])
count += 1
cpuTime = time.clock() - cpuStart
cConn.send((quals, cpuTime))
cConn.close()
def readLength(self, plot=False, printOut=True, pp=""):
"""
Analyzes the fastq file for the sequence length
Parameters
=======================
plot BOOL Produces a histogram plot of read lengths.
(Default: False)
printOut BOOL Prints statistics. (Default: True)
pp PdfPage Only used by the runQC function.
"""
if not self.inFile:
print "Needs an input file"
return
lens = []
count = 0
for seq, qual in cFileGen(self.inFile):
lens.append(len(seq))
count += 1
maxLen = max(lens)
minLen = min(lens)
if printOut:
print "Finished reading %i reads" % (count)
print "Min read length: %i" % (minLen)
print "Max read length: %i" % (maxLen)
if plot:
plt.figure()
plt.hist(lens)
plt.title("Histogram of Read Lengths")
plt.ylabel("Counts")
plt.xlabel("Read Length")
if pp:
pp.savefig()
else:
plt.show()
self.maxLen = maxLen
self.numReads = count
def plotQual(self, printOut=True, nCores=1, verbose=False, plot=True, pp=""):
"""
View a boxplot of the qualities by base.
Parameters
=======================
printOut BOOL print quality format (Default: True)
nCores INT Number of cpu cores to use. -1 to use all cores. (Default: 1)
verbose BOOL Print the runtime (Default: False)
pp PdfPage Only used by the runQC function.
"""
if not "self.maxLen" in locals():
self.readLength(printOut=False)
quals = initMatrix(self.maxLen)
if nCores == 1:
cpuStart = time.clock()
wallStart = time.time()
for seq, qual in cFileGen(self.inFile):
tmp = map(ord, qual)
for i in xrange(len(tmp)):
quals[i].append(tmp[i])
cpuTotal = time.clock() - cpuStart
wallTime = time.time() - wallStart
else:
nCores = setCores(nCores, verbose)
p = [] # process array
pConns = [] # parent connection array
for i in xrange(nCores):
pConn, cConn = Pipe() # returns (parent connection, child connection)
pConns.append(pConn)
p.append(
Process(target=qualWorker, args=(self.inFile, self.maxLen, i, nCores, cConn))
) # initialize processes
wallStart = time.time()
for i in xrange(nCores): # start processes
p[i].start()
cpuTotal = 0
for i in xrange(nCores):
tmpQuals, cpuTime = pConns[i].recv() # get results from processes
for i in xrange(len(tmpQuals)):
quals[i].extend(tmpQuals[i])
cpuTotal += cpuTime
wallTime = time.time() - wallStart
for i in xrange(nCores):
p[i].join()
print sum(map(sum, quals))
if verbose:
print "CPU time: %.3f seconds" % (cpuTotal)
print "Walltime: %.3f seconds" % (wallTime)
qualRange = calcQualRange(quals)
self.qualRange = qualRange
if printOut:
print "Quality format: +%d" % (qualRange[0])
if plot:
plt.figure(figsize=(18, 3))
plt.boxplot(quals, sym="")
plt.plot(range(1, len(quals) + 1), map(np.mean, quals))
plt.title("%s Quality Plot" % (self.inFile.split("/")[-1]))
plt.ylabel("Quality Score")
plt.ylim(qualRange)
plt.tick_params(axis="x", which="both", labelbottom="off")
plt.tight_layout()
if pp:
pp.savefig()
else:
plt.show()
def calcKmers(self, k=6, nCores=1, plot=True, adapt=False, assemble=True, verbose=False, pp=""):
"""
Calculate and make a scatter plot of k-mers in reads.
Parameters
======================
k INT k-mer size (Default: 5)
nCores INT cores to use (Default: 1)
plot BOOL plot the output (Default: True)
adapt BOOL align kmers against illumina adapters (Default:True)
verbose BOOL print the runtime (Default: False)
pp PdfPage Only used by the runQC function.
"""
kmerDict = Counter()
if nCores == 1:
wallStart = time.time()
cpuStart = time.clock()
for seq, qual in cFileGen(self.inFile):
tmp = [seq[i : i + k] for i in xrange(0, len(seq) - (k - 1), 3)]
for i in tmp:
if not "N" in i:
kmerDict[i] += 1
wallTime = time.time() - wallStart
cpuTotal = time.clock()
else:
nCores = setCores(nCores, verbose)
p = [] # process array
pConns = [] # parent connection array
for i in xrange(nCores):
pConn, cConn = Pipe() # returns (parent connection, child connection)
pConns.append(pConn)
# initialize processes
p.append(Process(target=kmerWorker, args=(self.inFile, k, i, nCores, cConn)))
wallStart = time.time()
for i in xrange(nCores): # start processes
p[i].start()
cpuTotal = 0
for i in xrange(nCores):
kmerTop100, cpuTime = pConns[i].recv() # get results from processes
cpuTotal += cpuTime
for kmer, v in kmerTop100:
kmerDict[kmer] += v
wallTime = time.time() - wallStart
for i in xrange(nCores):
p[i].join()
if verbose:
print "CPU time: %.3f seconds" % (cpuTotal)
print "Walltime: %.3f seconds" % (wallTime)
top20 = kmerDict.most_common(20)
if plot:
vals = map(lambda y: y[1], top20)
bottoms = np.cumsum([0] + vals[:-1])
plt.figure(figsize=(4, 8))
plt.axis("off")
plt.bar(np.zeros(20), vals, width=np.ones(20), color=cm.Set1(np.linspace(0, 1, 20)), bottom=bottoms)
plt.xlim((-0.05, 1.6))
for i in xrange(20):
plt.text(1.1, bottoms[i] + vals[i] / 2.0, top20[i][0], verticalalignment="center", family="monospace")
plt.text(0.5, bottoms[i] + vals[i] / 2.0, str(vals[i]), va="center", ha="center")
fName = self.inFile.split("/")[-1]
plt.title("Top 20 K-mers in " + fName)
plt.tight_layout()
if pp:
pp.savefig()
else:
plt.show()
if adapt:
print "\nAdapter Alignment to Kmers\n=============================="
adaptAlign(map(lambda y: y[0], top20), nCores)
if assemble:
newKmers = map(lambda y: y[0], kmerDict.most_common(25))
kmers = []
while len(kmers) != len(newKmers) or change:
change = False
kmers = newKmers
newKmers = []
kept = [False for i in range(len(kmers))]
for i in range(len(kmers) - 1):
iLen = len(kmers[i])
for j in range(i + 1, len(kmers)):
jLen = len(kmers[j])
minLen = min((iLen, jLen))
alignRes = pairwise2.align.localms(kmers[i], kmers[j], 2, -2, -2, -2)
if alignRes:
if alignRes[0][2] == (minLen - 1) * 2.0:
change = True
new = pileup(alignRes[0][0], alignRes[0][1])
newKmers.append(new)
if change:
kept[i] = True
kept[j] = True
break
if change:
break
for i in range(len(kept)):
if not kept[i]:
newKmers.append(kmers[i])
print "\nAssembled Kmers\n=============================="
for i in kmers:
print i
print "\nAdapter Alignment to Assemblies\n=============================="
adaptAlign(kmers, nCores, thresh=0.9)
