def __init__(self,
insert_dist,
upper,
fasta,
pwm,
sep=120,
min_occ=0.1,
flank=60,
out=None,
bam=None,
ci=0.9,
step=5):
self.sep = sep
self.chrs = read_chrom_sizes_from_fasta(fasta)
self.fasta = fasta
if fasta is not None:
self.pwm = PWM.open(pwm)
self.window = flank * 2 + 1
self.min_occ = min_occ
self.flank = flank
self.bam = bam
self.upper = upper
self.occ_calc_params = OccupancyCalcParams(0,
upper,
insert_dist,
ci=ci)
if step % 2 == 0:
step = step - 1
self.step = step
self.halfstep = (self.step - 1) // 2
def run_nfr(args):
"""run nfr calling
"""
if args.bam is None and args.ins_track is None:
raise Exception("Must supply either bam file or insertion track")
if not args.out:
args.out = '.'.join(os.path.basename(args.calls).split('.')[0:-3])
if args.fasta is not None:
chrs_fasta = read_chrom_sizes_from_fasta(args.fasta)
pwm = PWM.open(args.pwm)
chunks = ChunkList.read(args.bed, chromDict = chrs_fasta, min_offset = max(pwm.up, pwm.down))
else:
chunks = ChunkList.read(args.bed)
if args.bam is not None:
chrs_bam = read_chrom_sizes_from_bam(args.bam)
chunks.checkChroms(chrs_bam, chrom_source = "BAM file")
chunks.merge()
maxQueueSize = args.cores * 10
params = NFRParameters(args.occ_track, args.calls, args.ins_track, args.bam, max_occ = args.max_occ, max_occ_upper = args.max_occ_upper,
fasta = args.fasta, pwm = args.pwm)
sets = chunks.split(items = args.cores * 5)
pool1 = mp.Pool(processes = max(1,args.cores-1))
nfr_handle = open(args.out + '.nfrpos.bed','w')
nfr_handle.close()
nfr_queue = mp.JoinableQueue()
nfr_process = mp.Process(target = _writeNFR, args=(nfr_queue, args.out))
nfr_process.start()
if params.ins_track is None:
ins_handle = open(args.out + '.ins.bedgraph','w')
ins_handle.close()
ins_queue = mp.JoinableQueue()
ins_process = mp.Process(target = _writeIns, args=(ins_queue, args.out))
ins_process.start()
for j in sets:
tmp = pool1.map(_nfrHelper, zip(j,itertools.repeat(params)))
for result in tmp:
if params.ins_track is None:
nfr_queue.put(result[0])
ins_queue.put(result[1])
else:
nfr_queue.put(result)
pool1.close()
pool1.join()
nfr_queue.put('STOP')
nfr_process.join()
if params.ins_track is None:
ins_queue.put('STOP')
ins_process.join()
pysam.tabix_compress(args.out + '.nfrpos.bed', args.out + '.nfrpos.bed.gz',force = True)
shell_command('rm ' + args.out + '.nfrpos.bed')
pysam.tabix_index(args.out + '.nfrpos.bed.gz', preset = "bed", force = True)
if params.ins_track is None:
pysam.tabix_compress(args.out + '.ins.bedgraph', args.out + '.ins.bedgraph.gz', force = True)
shell_command('rm ' + args.out + '.ins.bedgraph')
pysam.tabix_index(args.out + '.ins.bedgraph.gz', preset = "bed", force = True)
def __init__(self, flank, lower, upper, bg, fasta, pwm, sizes, scale):
self.flank = flank
self.lower = lower
self.upper = upper
self.scale = scale
self.bg = bg
self.fasta = fasta
if self.bg is None:
self.pwm = PWM.open(pwm)
self.chrs = read_chrom_sizes_from_fasta(fasta)
self.fragmentsizes = FragmentSizes.open(sizes)
def __init__(self, flank, lower, upper, bg, fasta, pwm, sizes, scale):
self.flank = flank
self.lower = lower
self.upper = upper
self.scale = scale
self.bg = bg
self.fasta = fasta
if self.bg is None:
self.pwm = PWM.open(pwm)
self.chrs = read_chrom_sizes_from_fasta(fasta)
self.fragmentsizes = FragmentSizes.open(sizes)
def __init__(self, occ_track,calls, ins_track = None, bam = None, max_occ = 0.25, max_occ_upper = 0.25, fasta = None, pwm = None):
self.bam = bam
self.ins_track = ins_track
self.occ_track = occ_track
self.calls = calls
self.max_occ = max_occ
self.max_occ_upper = max_occ_upper
self.fasta = fasta
if fasta is not None:
self.pwm = PWM.open(pwm)
self.chrs = read_chrom_sizes_from_fasta(fasta)
def get_pwm(args, bases=50000, splitsize=1000):
"""Functiono obtain PWM around ATAC insertion"""
if not args.out:
args.out = '.'.join(os.path.basename(args.bam).split('.')[0:-1])
chrs = read_chrom_sizes_from_fasta(args.fasta)
if args.bed is None:
chunks = ChunkList.convertChromSizes(chrs,
splitsize=splitsize,
offset=args.flank)
sets = chunks.split(items=bases / splitsize)
else:
chunks = ChunkList.read(args.bed,
chromDict=chrs,
min_offset=args.flank)
sets = chunks.split(bases=bases)
params = _PWMParameters(bam=args.bam,
up=args.flank,
down=args.flank,
fasta=args.fasta,
lower=args.lower,
upper=args.upper,
atac=args.atac,
sym=args.sym)
pool = Pool(processes=args.cores)
tmp = pool.map(_pwmHelper, zip(sets, itertools.repeat(params)))
pool.close()
pool.join()
n = 0.0
result = np.zeros((len(params.nucleotides), params.up + params.down + 1))
for i in tmp:
result += i[0]
n += i[1]
result /= n
if args.bed:
normfreqs = seq.getNucFreqsFromChunkList(chunks, args.fasta,
params.nucleotides)
else:
normfreqs = seq.getNucFreqs(args.fasta, params.nucleotides)
result = result / np.reshape(np.repeat(normfreqs, result.shape[1]),
result.shape)
if args.sym:
#Symmetrize
left = result[:, 0:(args.flank + 1)]
right = result[:, args.flank:]
rightflipped = np.fliplr(np.flipud(right))
combined = (left + rightflipped) / 2
result = np.hstack(
(combined, np.fliplr(np.flipud(combined[:, 0:args.flank]))))
#save
pwm = PWM(result, args.flank, args.flank, params.nucleotides)
pwm.save(args.out + '.PWM.txt')
def __init__(self, up, down, fasta, dinucleotide = False):
self.up = up
self.down = down
self.fasta = fasta
self.chrs = utils.read_chrom_sizes_from_fasta(fasta)
if dinucleotide:
nucs="CGAT"
dinucs = []
for p in itertools.product(nucs, repeat=2):
dinucs.append("".join(p))
self.nucleotides = dinucs
else:
self.nucleotides = ["A","C","G","T"]
self.matsize = (len(self.nucleotides), self.up + self.down + 1)
self.dinucleotide = dinucleotide
def __init__(self, up, down, fasta, dinucleotide=False):
self.up = up
self.down = down
self.fasta = fasta
self.chrs = utils.read_chrom_sizes_from_fasta(fasta)
if dinucleotide:
nucs = "CGAT"
dinucs = []
for p in itertools.product(nucs, repeat=2):
dinucs.append("".join(p))
self.nucleotides = dinucs
else:
self.nucleotides = ["A", "C", "G", "T"]
self.matsize = (len(self.nucleotides), self.up + self.down + 1)
self.dinucleotide = dinucleotide
def __init__(self, insert_dist, upper, fasta, pwm, sep = 120, min_occ = 0.1, flank = 60,
out = None, bam = None, ci = 0.9, step = 5):
self.sep = sep
self.chrs = read_chrom_sizes_from_fasta(fasta)
self.fasta = fasta
if fasta is not None:
self.pwm = PWM.open(pwm)
self.window = flank * 2 + 1
self.min_occ = min_occ
self.flank = flank
self.bam = bam
self.upper = upper
self.occ_calc_params = OccupancyCalcParams(0, upper, insert_dist, ci = ci)
if step%2 == 0:
step = step - 1
self.step = step
self.halfstep = (self.step-1) / 2
def get_pwm(args, bases = 50000, splitsize = 1000):
"""Functiono obtain PWM around ATAC insertion"""
if not args.out:
args.out = '.'.join(os.path.basename(args.bam).split('.')[0:-1])
chrs = read_chrom_sizes_from_fasta(args.fasta)
if args.bed is None:
chunks = ChunkList.convertChromSizes(chrs, splitsize = splitsize, offset = args.flank)
sets = chunks.split(items = bases/splitsize)
else:
chunks = ChunkList.read(args.bed, chromDict = chrs, min_offset = args.flank)
sets = chunks.split(bases = bases)
params = _PWMParameters(bam = args.bam, up = args.flank, down = args.flank, fasta = args.fasta,
lower = args.lower, upper = args.upper, atac = args.atac, sym = args.sym)
pool = Pool(processes = args.cores)
tmp = pool.map(_pwmHelper, zip(sets,itertools.repeat(params)))
pool.close()
pool.join()
n = 0.0
result = np.zeros((len(params.nucleotides), params.up + params.down + 1))
for i in tmp:
result += i[0]
n += i[1]
result /= n
if args.bed:
normfreqs = seq.getNucFreqsFromChunkList(chunks, args.fasta, params.nucleotides)
else:
normfreqs = seq.getNucFreqs(args.fasta, params.nucleotides)
result = result / np.reshape(np.repeat(normfreqs,result.shape[1]),result.shape)
if args.sym:
#Symmetrize
left = result[:,0:(args.flank + 1)]
right = result[:,args.flank:]
rightflipped = np.fliplr(np.flipud(right))
combined = (left + rightflipped) / 2
result = np.hstack((combined, np.fliplr(np.flipud(combined[:,0:args.flank]))))
#save
pwm = PWM(result, args.flank, args.flank, params.nucleotides)
pwm.save(args.out + '.PWM.txt')
def run_occ(args):
"""run occupancy calling
"""
if args.fasta:
chrs = read_chrom_sizes_from_fasta(args.fasta)
else:
chrs = read_chrom_sizes_from_bam(args.bam)
pwm = PWM.open(args.pwm)
chunks = ChunkList.read(args.bed, chromDict = chrs, min_offset = args.flank + args.upper/2 + max(pwm.up,pwm.down) + args.nuc_sep/2)
chunks.slop(chrs, up = args.nuc_sep/2, down = args.nuc_sep/2)
chunks.merge()
maxQueueSize = args.cores*10
fragment_dist = FragmentMixDistribution(0, upper = args.upper)
if args.sizes is not None:
tmp = FragmentSizes.open(args.sizes)
fragment_dist.fragmentsizes = FragmentSizes(0, args.upper, vals = tmp.get(0,args.upper))
else:
fragment_dist.getFragmentSizes(args.bam, chunks)
fragment_dist.modelNFR()
fragment_dist.plotFits(args.out + '.occ_fit.eps')
fragment_dist.fragmentsizes.save(args.out + '.fragmentsizes.txt')
params = OccupancyParameters(fragment_dist, args.upper, args.fasta, args.pwm, sep = args.nuc_sep, min_occ = args.min_occ,
flank = args.flank, bam = args.bam, ci = args.confidence_interval, step = args.step)
sets = chunks.split(items = args.cores * 5)
pool1 = mp.Pool(processes = max(1,args.cores-1))
out_handle1 = open(args.out + '.occ.bedgraph','w')
out_handle1.close()
out_handle2 = open(args.out + '.occ.lower_bound.bedgraph','w')
out_handle2.close()
out_handle3 = open(args.out + '.occ.upper_bound.bedgraph','w')
out_handle3.close()
write_queue = mp.JoinableQueue(maxsize = maxQueueSize)
write_process = mp.Process(target = _writeOcc, args=(write_queue, args.out))
write_process.start()
peaks_handle = open(args.out + '.occpeaks.bed','w')
peaks_handle.close()
peaks_queue = mp.JoinableQueue()
peaks_process = mp.Process(target = _writePeaks, args=(peaks_queue, args.out))
peaks_process.start()
nuc_dist = np.zeros(args.upper)
for j in sets:
tmp = pool1.map(_occHelper, zip(j,itertools.repeat(params)))
for result in tmp:
nuc_dist += result[0]
write_queue.put(result[1])
peaks_queue.put(result[2])
pool1.close()
pool1.join()
write_queue.put('STOP')
peaks_queue.put('STOP')
write_process.join()
peaks_process.join()
pysam.tabix_compress(args.out + '.occpeaks.bed', args.out + '.occpeaks.bed.gz',force = True)
shell_command('rm ' + args.out + '.occpeaks.bed')
pysam.tabix_index(args.out + '.occpeaks.bed.gz', preset = "bed", force = True)
for i in ('occ','occ.lower_bound','occ.upper_bound'):
pysam.tabix_compress(args.out + '.' + i + '.bedgraph', args.out + '.'+i+'.bedgraph.gz',force = True)
shell_command('rm ' + args.out + '.' + i + '.bedgraph')
pysam.tabix_index(args.out + '.' + i + '.bedgraph.gz', preset = "bed", force = True)
dist_out = FragmentSizes(0, args.upper, vals = nuc_dist)
dist_out.save(args.out + '.nuc_dist.txt')
print "Making figure"
#make figure
fig = plt.figure()
plt.plot(range(0,args.upper),dist_out.get(0,args.upper),label = "Nucleosome Distribution")
plt.xlabel("Fragment Size")
plt.ylabel("Frequency")
fig.savefig(args.out+'.nuc_dist.eps')
plt.close(fig)
def run_nuc(args):
"""run occupancy calling
"""
vmat = VMat.open(args.vmat)
if args.fasta:
chrs = read_chrom_sizes_from_fasta(args.fasta)
else:
chrs = read_chrom_sizes_from_bam(args.bam)
pwm = PWM.open(args.pwm)
chunks = ChunkList.read(args.bed,
chromDict=chrs,
min_offset=vmat.mat.shape[1] + vmat.upper // 2 +
max(pwm.up, pwm.down) + args.nuc_sep // 2,
min_length=args.nuc_sep * 2)
chunks.slop(chrs, up=args.nuc_sep // 2, down=args.nuc_sep // 2)
chunks.merge()
maxQueueSize = args.cores * 10
if args.sizes is not None:
fragment_dist = FragmentSizes.open(args.sizes)
else:
fragment_dist = FragmentSizes(0, upper=vmat.upper)
fragment_dist.calculateSizes(args.bam, chunks)
params = NucParameters(vmat=vmat,
fragmentsizes=fragment_dist,
bam=args.bam,
fasta=args.fasta,
pwm=args.pwm,
occ_track=args.occ_track,
sd=args.sd,
nonredundant_sep=args.nuc_sep,
redundant_sep=args.redundant_sep,
min_z=args.min_z,
min_lr=args.min_lr,
atac=args.atac)
sets = chunks.split(items=args.cores * 5)
pool1 = mp.Pool(processes=max(1, args.cores - 1))
if args.write_all:
outputs = [
'nucpos', 'nucpos.redundant', 'nucleoatac_signal',
'nucleoatac_signal.smooth', 'nucleoatac_background',
'nucleoatac_raw'
]
else:
outputs = [
'nucpos', 'nucpos.redundant', 'nucleoatac_signal',
'nucleoatac_signal.smooth'
]
handles = {}
write_queues = {}
write_processes = {}
for i in outputs:
if i not in ['nucpos', 'nucpos.redundant', 'nfrpos']:
handles[i] = open(args.out + '.' + i + '.bedgraph', 'w')
else:
handles[i] = open(args.out + '.' + i + '.bed', 'w')
handles[i].close()
write_queues[i] = mp.JoinableQueue(maxsize=maxQueueSize)
write_processes[i] = mp.Process(target=_writeFuncs[i],
args=(write_queues[i], args.out))
write_processes[i].start()
for j in sets:
tmp = pool1.map(_nucHelper, list(zip(j, itertools.repeat(params))))
for result in tmp:
for i in outputs:
write_queues[i].put(result[i])
pool1.close()
pool1.join()
for i in outputs:
write_queues[i].put('STOP')
for i in outputs:
write_processes[i].join()
if i not in ['nucpos', 'nucpos.redundant']:
pysam.tabix_compress(args.out + '.' + i + '.bedgraph',
args.out + '.' + i + '.bedgraph.gz',
force=True)
shell_command('rm ' + args.out + '.' + i + '.bedgraph')
pysam.tabix_index(args.out + '.' + i + '.bedgraph.gz',
preset="bed",
force=True)
else:
pysam.tabix_compress(args.out + '.' + i + '.bed',
args.out + '.' + i + '.bed.gz',
force=True)
shell_command('rm ' + args.out + '.' + i + '.bed')
pysam.tabix_index(args.out + '.' + i + '.bed.gz',
preset="bed",
force=True)
def __init__(self, fasta, pwm):
self.chrs = read_chrom_sizes_from_fasta(fasta)
self.fasta = fasta
self.pwm = PWM.open(pwm)
def __init__(self, fasta, pwm):
self.chrs = read_chrom_sizes_from_fasta(fasta)
self.fasta = fasta
self.pwm = PWM.open(pwm)
def run_nfr(args):
"""run nfr calling
"""
if args.bam is None and args.ins_track is None:
raise Exception("Must supply either bam file or insertion track")
if not args.out:
args.out = '.'.join(os.path.basename(args.calls).split('.')[0:-3])
if args.fasta is not None:
chrs_fasta = read_chrom_sizes_from_fasta(args.fasta)
pwm = PWM.open(args.pwm)
chunks = ChunkList.read(args.bed,
chromDict=chrs_fasta,
min_offset=max(pwm.up, pwm.down))
else:
chunks = ChunkList.read(args.bed)
if args.bam is not None:
chrs_bam = read_chrom_sizes_from_bam(args.bam)
chunks.checkChroms(chrs_bam, chrom_source="BAM file")
chunks.merge()
maxQueueSize = args.cores * 10
params = NFRParameters(args.occ_track,
args.calls,
args.ins_track,
args.bam,
max_occ=args.max_occ,
max_occ_upper=args.max_occ_upper,
fasta=args.fasta,
pwm=args.pwm)
sets = chunks.split(items=args.cores * 5)
pool1 = mp.Pool(processes=max(1, args.cores - 1))
nfr_handle = open(args.out + '.nfrpos.bed', 'w')
nfr_handle.close()
nfr_queue = mp.JoinableQueue()
nfr_process = mp.Process(target=_writeNFR, args=(nfr_queue, args.out))
nfr_process.start()
if params.ins_track is None:
ins_handle = open(args.out + '.ins.bedgraph', 'w')
ins_handle.close()
ins_queue = mp.JoinableQueue()
ins_process = mp.Process(target=_writeIns, args=(ins_queue, args.out))
ins_process.start()
for j in sets:
tmp = pool1.map(_nfrHelper, zip(j, itertools.repeat(params)))
for result in tmp:
if params.ins_track is None:
nfr_queue.put(result[0])
ins_queue.put(result[1])
else:
nfr_queue.put(result)
pool1.close()
pool1.join()
nfr_queue.put('STOP')
nfr_process.join()
if params.ins_track is None:
ins_queue.put('STOP')
ins_process.join()
pysam.tabix_compress(args.out + '.nfrpos.bed',
args.out + '.nfrpos.bed.gz',
force=True)
shell_command('rm ' + args.out + '.nfrpos.bed')
pysam.tabix_index(args.out + '.nfrpos.bed.gz', preset="bed", force=True)
if params.ins_track is None:
pysam.tabix_compress(args.out + '.ins.bedgraph',
args.out + '.ins.bedgraph.gz',
force=True)
shell_command('rm ' + args.out + '.ins.bedgraph')
pysam.tabix_index(args.out + '.ins.bedgraph.gz',
preset="bed",
force=True)
def run_nuc(args):
"""run occupancy calling
"""
vmat = VMat.open(args.vmat)
if args.fasta:
chrs = read_chrom_sizes_from_fasta(args.fasta)
else:
chrs = read_chrom_sizes_from_bam(args.bam)
pwm = PWM.open(args.pwm)
chunks = ChunkList.read(args.bed, chromDict = chrs, min_offset = vmat.mat.shape[1] + vmat.upper/2 + max(pwm.up,pwm.down) + args.nuc_sep/2, min_length = args.nuc_sep * 2)
chunks.slop(chrs, up = args.nuc_sep/2, down = args.nuc_sep/2)
chunks.merge()
maxQueueSize = args.cores*10
if args.sizes is not None:
fragment_dist = FragmentSizes.open(args.sizes)
else:
fragment_dist = FragmentSizes(0, upper = vmat.upper)
fragment_dist.calculateSizes(args.bam, chunks)
params = NucParameters(vmat = vmat, fragmentsizes = fragment_dist, bam = args.bam, fasta = args.fasta, pwm = args.pwm,
occ_track = args.occ_track,
sd = args.sd, nonredundant_sep = args.nuc_sep, redundant_sep = args.redundant_sep,
min_z = args.min_z, min_lr = args.min_lr , atac = args.atac)
sets = chunks.split(items = args.cores*5)
pool1 = mp.Pool(processes = max(1,args.cores-1))
if args.write_all:
outputs = ['nucpos','nucpos.redundant','nucleoatac_signal','nucleoatac_signal.smooth',
'nucleoatac_background','nucleoatac_raw']
else:
outputs = ['nucpos','nucpos.redundant','nucleoatac_signal','nucleoatac_signal.smooth']
handles = {}
write_queues = {}
write_processes = {}
for i in outputs:
if i not in ['nucpos','nucpos.redundant','nfrpos']:
handles[i] = open(args.out + '.'+i+'.bedgraph','w')
else:
handles[i] = open(args.out + '.'+i+'.bed','w')
handles[i].close()
write_queues[i] = mp.JoinableQueue(maxsize = maxQueueSize)
write_processes[i] = mp.Process(target = _writeFuncs[i], args=(write_queues[i], args.out))
write_processes[i].start()
for j in sets:
tmp = pool1.map(_nucHelper, zip(j,itertools.repeat(params)))
for result in tmp:
for i in outputs:
write_queues[i].put(result[i])
pool1.close()
pool1.join()
for i in outputs:
write_queues[i].put('STOP')
for i in outputs:
write_processes[i].join()
if i not in ['nucpos','nucpos.redundant']:
pysam.tabix_compress(args.out + '.' + i + '.bedgraph', args.out + '.' + i + '.bedgraph.gz',force = True)
shell_command('rm ' + args.out + '.' + i + '.bedgraph')
pysam.tabix_index(args.out + '.' + i + '.bedgraph.gz', preset = "bed", force = True)
else:
pysam.tabix_compress(args.out + '.' + i + '.bed', args.out + '.' + i + '.bed.gz',force = True)
shell_command('rm ' + args.out + '.' + i + '.bed')
pysam.tabix_index(args.out + '.' + i + '.bed.gz', preset = "bed", force = True)
