def __init__(self, normal_bam_filename, tumor_bam_filename,
reference_genome_filename, input_filename_base, segments_bed,
min_depth=20, min_bqual=10, min_mqual=10, process_num=1):
self.normal_bam_filename = normal_bam_filename
self.tumor_bam_filename = tumor_bam_filename
self.reference_genome_filename = reference_genome_filename
self.input_filename_base = input_filename_base
self.segments_bed = segments_bed
self.min_depth = min_depth
self.min_bqual = min_bqual
self.min_mqual = min_mqual
self.process_num = process_num
self.data = Data()
def __init__(self,
normal_bam_filename,
tumor_bam_filename,
reference_genome_filename,
input_filename_base,
segments_bed,
BICseq_bed_fileName_corrected,
pkl_path="",
max_copynumber=6,
subclone_num=1,
baseline_thred_LOH=0.3,
baseline_thred_APM=0.01,
min_depth=20,
min_bqual=10,
min_mqual=10,
process_num=1):
self.normal_bam_filename = normal_bam_filename
self.tumor_bam_filename = tumor_bam_filename
self.reference_genome_filename = reference_genome_filename
self.input_filename_base = input_filename_base
self.segments_bed = segments_bed
self.BICseq_bed_fileName_corrected = BICseq_bed_fileName_corrected
self.pkl_path = pkl_path
self.max_copynumber = max_copynumber
self.subclone_num = subclone_num
self.baseline_thred_LOH = baseline_thred_LOH
self.baseline_thred_APM = baseline_thred_APM
self.min_depth = min_depth
self.min_bqual = min_bqual
self.min_mqual = min_mqual
print "process_num = {}".format(process_num)
self.process_num = process_num
self.data = Data()
class MixClone_Converter:
def __init__(self,
normal_bam_filename,
tumor_bam_filename,
reference_genome_filename,
input_filename_base,
segments_bed,
BICseq_bed_fileName_corrected,
pkl_path="",
max_copynumber=6,
subclone_num=1,
baseline_thred_LOH=0.3,
baseline_thred_APM=0.01,
min_depth=20,
min_bqual=10,
min_mqual=10,
process_num=1):
self.normal_bam_filename = normal_bam_filename
self.tumor_bam_filename = tumor_bam_filename
self.reference_genome_filename = reference_genome_filename
self.input_filename_base = input_filename_base
self.segments_bed = segments_bed
self.BICseq_bed_fileName_corrected = BICseq_bed_fileName_corrected
self.pkl_path = pkl_path
self.max_copynumber = max_copynumber
self.subclone_num = subclone_num
self.baseline_thred_LOH = baseline_thred_LOH
self.baseline_thred_APM = baseline_thred_APM
self.min_depth = min_depth
self.min_bqual = min_bqual
self.min_mqual = min_mqual
print "process_num = {}".format(process_num)
self.process_num = process_num
self.data = Data()
def convert(self, method, pkl_flag=False):
if pkl_flag and self.pkl_path != "":
print "load pkl from"
print self.pkl_path
infile = open(self.pkl_path, 'rb')
self.data = pkl.load(infile)
infile.close()
else:
self._load_segments()
print "MixClone converter converting"
if "auto" == method:
self._MCMC_gccorrection()
elif "visual" == method:
self._visual_gccorrection()
sys.stdout.flush()
self._get_counts()
self._output()
self._baseline_selection()
data_file_name = self.input_filename_base + '.MixClone.input.pkl'
outfile = open(data_file_name, 'wb')
pkl.dump(self.data, outfile, protocol=2)
outfile.close()
def _MCMC_gccorrection(self):
"""
The interception is irrelevant for correction, set as median
MCMCLM only returns the m and c, then correct the data here
"""
mcmclm = MCMCLM(self.data, 0, self.subclone_num, self.max_copynumber)
m, c = mcmclm.run()
print "MCMC slope = {}".format(m)
self._correct(m, c)
def _correct(self, slope, intercept):
x = np.array(map(lambda seg: seg.gc, self.data.segments))
y = np.array(
map(
lambda seg: np.log(seg.tumor_reads_num + 1) - np.log(
seg.normal_reads_num + 1), self.data.segments))
K = np.percentile(y, 50)
A = slope * x + intercept
y_corrected = y - A + K
for i in range(len(y_corrected)):
self.data.segments[i].tumor_reads_num = np.exp(
y_corrected[i] +
np.log(self.data.segments[i].normal_reads_num + 1)) - 1
print "gc corrected, with slope = {0}, intercept = {1}".\
format(slope, intercept)
def _visual_gccorrection(self):
gsp = GCStripePlot(self.data.segments, self.sampleNumber)
print "total number: {}".format(self.data.seg_num)
# Sampling then linear regression, poor performance
# gsp.sampleln([i * 1000 for i in range(1,9)], 100)
gsp.plot()
# todo trimed x, y position
x, y, m, c = gsp.output()
print "x, y, m, c"
print x, y, m, c
self._correct(m, c)
def _baseline_selection(self):
print "begin baseline selection.."
self._get_LOH_frac()
self._get_LOH_status()
self._get_APM_frac()
self._get_APM_status()
self._compute_Lambda_S()
def _get_APM_status(self):
self.data.get_APM_status(self.baseline_thred_APM)
def _get_LOH_status(self):
self.data.get_LOH_status(self.baseline_thred_LOH,
flag_runpreprocess=True)
def _compute_Lambda_S(self):
print "begin compute lambda s .."
self.data.compute_Lambda_S_LOH(self.max_copynumber,
self.subclone_num,
flag_runpreprocess=True)
def _output(self):
"""Output the parameter for THetA
The Upper and Lower Boundaries for normal heuristic
The GC corrected interval_count_file
"""
interval_count_file = open(self.BICseq_bed_fileName_corrected, 'w')
interval_count_file.write(
"ID\tchrm\tstart\tend\ttumorCount\tnormalCount\tgc\n")
for i in range(len(self.data.segments)):
ID_i = self.data.segments[i].chrom_idx
chrm_i = self.data.segments[i].chrom_name
start_i = self.data.segments[i].start
end_i = self.data.segments[i].end
tumorCount_i = self.data.segments[i].tumor_reads_num
normalCount_i = self.data.segments[i].normal_reads_num
gc_i = self.data.segments[i].gc
interval_count_file.write(
"{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}\n".format(
ID_i, chrm_i, start_i, end_i, tumorCount_i, normalCount_i,
gc_i))
interval_count_file.close()
print "GC corrected interval file generated!"
sys.stdout.flush()
def _load_segmentsn(self):
"""
:returns: TODO
"""
normal_bam = pysam.Samfile(self.normal_bam_filename, 'rb')
tumor_bam = pysam.Samfile(self.tumor_bam_filename, 'rb')
print 'Loading normalized segments by {0}...'.format(self.segments_bed)
sys.stdout.flush()
self.data.load_segmentsn(normal_bam, tumor_bam, self.segments_bed)
normal_bam.close()
tumor_bam.close()
def _load_segments(self):
normal_bam = pysam.Samfile(self.normal_bam_filename, 'rb')
tumor_bam = pysam.Samfile(self.tumor_bam_filename, 'rb')
print 'Loading segments with gc by {0}...'.format(self.segments_bed)
sys.stdout.flush()
# self.data.load_segments(normal_bam, tumor_bam, self.segments_bed)
self.data.load_segmentsn(self.segments_bed)
normal_bam.close()
tumor_bam.close()
def _get_counts(self):
seg_num = self.data.seg_num
process_num = self.process_num
print "process_num = {}".format(process_num)
if process_num > seg_num:
process_num = seg_num
pool = Pool(processes=process_num)
args_list = []
for j in range(0, seg_num):
seg_name = self.data.segments[j].name
chrom_name = self.data.segments[j].chrom_name
chrom_idx = self.data.segments[j].chrom_idx
start = self.data.segments[j].start
end = self.data.segments[j].end
args_tuple = (seg_name, chrom_name, chrom_idx, start, end,
self.normal_bam_filename, self.tumor_bam_filename,
self.reference_genome_filename, self.min_depth,
self.min_bqual, self.min_mqual)
args_list.append(args_tuple)
counts_tuple_list = pool.map(process_by_segment, args_list)
for j in range(0, seg_num):
paired_counts_j, BAF_counts_j = counts_tuple_list[j]
self.data.segments[j].paired_counts = paired_counts_j
self.data.segments[j].BAF_counts = BAF_counts_j
def _get_LOH_frac(self):
self.data.get_LOH_frac()
def _get_APM_frac(self):
self.data.get_APM_frac()
class BamToDataConverter:
def __init__(self, normal_bam_filename, tumor_bam_filename,
reference_genome_filename, input_filename_base, segments_bed,
min_depth=20, min_bqual=10, min_mqual=10, process_num=1):
self.normal_bam_filename = normal_bam_filename
self.tumor_bam_filename = tumor_bam_filename
self.reference_genome_filename = reference_genome_filename
self.input_filename_base = input_filename_base
self.segments_bed = segments_bed
self.min_depth = min_depth
self.min_bqual = min_bqual
self.min_mqual = min_mqual
self.process_num = process_num
self.data = Data()
def convert(self):
self._load_segments()
self._get_counts()
self._get_LOH_frac()
data_file_name = self.input_filename_base + '.MixClone.input.pkl'
outfile = open(data_file_name, 'wb')
pkl.dump(self.data, outfile, protocol=2)
outfile.close()
def _load_segments(self):
normal_bam = pysam.Samfile(self.normal_bam_filename, 'rb')
tumor_bam = pysam.Samfile(self.tumor_bam_filename, 'rb')
print 'Loading segments by {0}...'.format(self.segments_bed)
sys.stdout.flush()
self.data.load_segments(normal_bam, tumor_bam, self.segments_bed)
normal_bam.close()
tumor_bam.close()
def _get_counts(self):
seg_num = self.data.seg_num
process_num = self.process_num
if process_num > seg_num:
process_num = seg_num
pool = Pool(processes = process_num)
args_list = []
for j in range(0, seg_num):
seg_name = self.data.segments[j].name
chrom_name = self.data.segments[j].chrom_name
chrom_idx = self.data.segments[j].chrom_idx
start = self.data.segments[j].start
end = self.data.segments[j].end
args_tuple = (seg_name, chrom_name, chrom_idx, start, end, self.normal_bam_filename,
self.tumor_bam_filename, self.reference_genome_filename,
self.min_depth, self.min_bqual, self.min_mqual)
args_list.append(args_tuple)
counts_tuple_list = pool.map(process_by_segment, args_list)
for j in range(0, seg_num):
paired_counts_j, BAF_counts_j = counts_tuple_list[j]
self.data.segments[j].paired_counts = paired_counts_j
self.data.segments[j].BAF_counts = BAF_counts_j
def _get_LOH_frac(self):
self.data.get_LOH_frac()
