def read_call_stats_file(self):
fields = ['contig', 'position', 'ref_allele', 'alt_allele', 'tumor_name', 'normal_name', 't_alt_count',
't_ref_count'
, 'n_alt_count', 'n_ref_count','t_ref_sum','t_alt_sum', 'failure_reasons', 'judgement']
fields_type = {'contig': str, 'position': np.int, 'ref_allele': str, 'alt_allele': str, 'tumor_name': str,
'normal_name': str,
't_alt_count': np.int, 't_ref_count': np.int, 'n_alt_count': np.int, 'n_ref_count': np.int,
't_ref_sum': np.int,'t_alt_sum':np.int,
'failure_reasons': str, 'judgement': str}
try:
self.call_stats_table = pd.read_csv(self.call_stats_file, '\t', index_col=False,
comment='#', usecols=fields, dtype=fields_type)
except (ValueError, LookupError):
print 'Error reading call stats skipping first two rows and trying again'
self.call_stats_table = pd.read_csv(self.call_stats_file, '\t', index_col=False,
comment='#', skiprows=2, usecols=fields, dtype=fields_type)
if type(self.call_stats_table['contig'][0]) == str:
self.call_stats_table['Chromosome'] = du.chr2num(np.array(self.call_stats_table['contig']))
else:
self.call_stats_table['Chromosome'] = np.array(self.call_stats_table['contig']) - 1
self.call_stats_table = self.call_stats_table[np.isfinite(self.call_stats_table['Chromosome'])]
self.call_stats_table['genomic_coord_x'] = du.hg19_to_linear_positions(
np.array(self.call_stats_table['Chromosome']), np.array(self.call_stats_table['position']))
self.n_calls_in = len(self.call_stats_table)
self.call_stats_table.reset_index(inplace=True, drop=True)
def read_call_stats_file(self):
try:
self.call_stats_table = pd.read_csv(self.call_stats_file,
'\t',
index_col=False,
low_memory=False,
comment='#')
except (ValueError, LookupError):
print 'Error reading call stats skipping first two rows and trying again'
self.call_stats_table = pd.read_csv(self.call_stats_file,
'\t',
index_col=False,
low_memory=False,
comment='#',
skiprows=2)
if type(self.call_stats_table['contig'][0]) == str:
self.call_stats_table['Chromosome'] = du.chr2num(
np.array(self.call_stats_table['contig']))
else:
self.call_stats_table['Chromosome'] = np.array(
self.call_stats_table['contig']) - 1
self.call_stats_table = self.call_stats_table[np.isfinite(
self.call_stats_table['Chromosome'])]
self.call_stats_table['genomic_coord_x'] = du.hg19_to_linear_positions(
np.array(self.call_stats_table['Chromosome']),
np.array(self.call_stats_table['position']))
self.n_calls_in = len(self.call_stats_table)
self.call_stats_table.reset_index(inplace=True, drop=True)
def read_het_file(self):
tumor_het_table = pd.read_csv(self.tumor_het_file,
'\t',
index_col=False,
low_memory=False,
comment='#')
normal_het_table = pd.read_csv(self.normal_het_file,
'\t',
index_col=False,
low_memory=False,
comment='#')
tumor_het_table = du.fix_het_file_header(tumor_het_table)
normal_het_table = du.fix_het_file_header(normal_het_table)
if type(tumor_het_table['CONTIG'][0]) == str:
tumor_het_table['Chromosome'] = du.chr2num(
np.array(tumor_het_table['CONTIG']))
else:
tumor_het_table['Chromosome'] = np.array(tumor_het_table['CONTIG'])
if type(normal_het_table['CONTIG'][0]) == str:
normal_het_table['Chromosome'] = du.chr2num(
np.array(normal_het_table['CONTIG']))
else:
normal_het_table['Chromosome'] = np.array(
normal_het_table['CONTIG'])
tumor_het_table = tumor_het_table[np.isfinite(
tumor_het_table['Chromosome'])]
tumor_het_table['genomic_coord_x'] = du.hg19_to_linear_positions(
np.array(tumor_het_table['Chromosome']),
np.array(tumor_het_table['POSITION']))
normal_het_table = normal_het_table[np.isfinite(
normal_het_table['Chromosome'])]
normal_het_table['genomic_coord_x'] = du.hg19_to_linear_positions(
np.array(normal_het_table['Chromosome']),
np.array(normal_het_table['POSITION']))
tumor_het_table['AF'] = np.true_divide(
tumor_het_table['ALT_COUNT'],
tumor_het_table['ALT_COUNT'] + tumor_het_table['REF_COUNT'])
normal_het_table['AF'] = np.true_divide(
normal_het_table['ALT_COUNT'],
normal_het_table['ALT_COUNT'] + normal_het_table['REF_COUNT'])
self.het_table = pd.merge(normal_het_table,
tumor_het_table,
on='genomic_coord_x',
suffixes=('_N', '_T'))
def read_seg_file(self):
if self.seg_file == 'NULL':
self.seg_table = pd.DataFrame(index=[0],columns=['Chromosome','Start.bp','End.bp','n_probes','length','f','tau','genomic_coord_start','genomic_coord_end'])
self.het_table = pd.DataFrame(index=[0],columns=['seg_id','tau','f','d','AF_T','AF_N','Chromosome','genomic_coord_x','ALT_COUNT_N'
'ALT_COUNT_T','REF_COUNT_N','REF_COUNT_T'])
else:
seg_header = du.read_file_header(self.seg_file)
cols_seg_type = {seg_header[0]: str}
self.seg_table = pd.read_csv(self.seg_file, '\t', index_col=False, low_memory=False, comment='#',
dtype=cols_seg_type)
self.seg_table = du.fix_seg_file_header(self.seg_table)
self.seg_table['Chromosome'] = du.chr2num(np.array(self.seg_table['Chromosome']))
self.seg_table['genomic_coord_start'] = du.hg19_to_linear_positions(np.array(self.seg_table['Chromosome']),
np.array(self.seg_table['Start.bp']))
self.seg_table['genomic_coord_end'] = du.hg19_to_linear_positions(np.array(self.seg_table['Chromosome']),
np.array(self.seg_table['End.bp']))
def read_seg_file(self):
self.seg_table = pd.read_csv(self.seg_file,
'\t',
index_col=False,
low_memory=False,
comment='#')
self.seg_table = du.fix_seg_file_header(self.seg_table)
if not du.is_number(self.seg_table['Chromosome'][0]):
self.seg_table['Chromosome'] = du.chr2num(
np.array(self.seg_table['Chromosome']))
else:
self.seg_table['Chromosome'] = self.seg_table['Chromosome'] - 1
self.seg_table['genomic_coord_start'] = du.hg19_to_linear_positions(
np.array(self.seg_table['Chromosome']),
np.array(self.seg_table['Start.bp']))
self.seg_table['genomic_coord_end'] = du.hg19_to_linear_positions(
np.array(self.seg_table['Chromosome']),
np.array(self.seg_table['End.bp']))
def __init__(self,
candidate_sites,
p_somatic,
resolution=101,
f_thresh=0.15,
depth=15,
hot_spots_file='NA',
skew=0.5):
# variables follow notation:
# ac = allele count n = normal t = tumor
# Variables for SSNV fit
self.TiN_range = np.linspace(0, 1, num=resolution)
self.af = np.linspace(0.005, 1, num=200)
# observed data
self.contig = candidate_sites['contig']
self.position = candidate_sites['position']
self.genomic_coord_x = candidate_sites['genomic_coord_x']
self.n_alt_count = np.array(candidate_sites['n_alt_count'])
self.n_ref_count = np.array(candidate_sites['n_ref_count'])
self.n_depth = self.n_alt_count + self.n_ref_count
self.normal_f = np.nan_to_num(
np.true_divide(self.n_alt_count, self.n_depth))
self.t_alt_count = np.array(candidate_sites['t_alt_count'])
self.t_ref_count = np.array(candidate_sites['t_ref_count'])
self.t_depth = self.t_alt_count + self.t_ref_count
self.tumor_f = np.true_divide(self.t_alt_count, self.t_depth)
self.number_of_sites = len(self.n_alt_count)
self.candidate_sites = np.logical_and(
np.logical_and(self.tumor_f > f_thresh, self.t_depth > depth),
self.n_depth > depth)
# hyperparameter
self.p_somatic = np.zeros([self.number_of_sites, 1]) + p_somatic
if hot_spots_file != 'NA':
hot_spots = pd.read_csv(hot_spots_file,
sep='\t',
low_memory=False,
index_col=False)
if type(hot_spots['Chromosome'][0]) == str:
hot_spots['contig'] = du.chr2num(
np.array(hot_spots['Chromosome']))
else:
hot_spots['contig'] = np.array(hot_spots['Chromosome']) - 1
hot_spots = hot_spots[np.isfinite(hot_spots['contig'])]
hot_spots['genomic_coord_x'] = du.hg19_to_linear_positions(
np.array(hot_spots['contig']), np.array(hot_spots['Position']))
for index, hot_spot in hot_spots.iterrows():
if np.size(
np.where(self.genomic_coord_x ==
hot_spot['genomic_coord_x'])) > 0:
print 'Using user provided probabilities for cancer hot spots:'
print hot_spot['Chromosome'] + ' ' + hot_spot['Position']
self.p_somatic[np.where(
self.genomic_coord_x ==
hot_spot['genomic_coord_x'])] = hot_spot['Probability']
# parameter
self.TiN = 0
self.CI_tin_high = []
self.CI_tin_low = []
self.E_z = np.zeros([self.number_of_sites, 1])
self.skew = skew
# expected allele fraction of minor allele given allelic copy data
self.psi = .5 - np.array(candidate_sites['f_acs'])
self.t_het_direction = self.tumor_f < self.skew
self.t_het_direction = self.t_het_direction * -1
self.t_het_direction[self.t_het_direction == 0] = 1
# determine ratio of tumor to normal copies given tau and TiN at each locus
self.tau = candidate_sites['tau']
self.tin_correct_tau = np.multiply(
self.TiN_range, candidate_sites['tau'][:, np.newaxis])
self.tin_correct_normal_tau = np.multiply((1 - self.TiN_range), 2)
self.CN_ratio = np.divide(
self.tin_correct_tau,
np.array(self.tin_correct_tau + self.tin_correct_normal_tau))
# random variables
self.rv_normal_af = beta(self.n_alt_count + 1, self.n_ref_count + 1)
self.rv_tumor_af = beta(self.t_alt_count + 1, self.t_ref_count + 1)
# conditionals
self.p_TiN_given_S = np.zeros([self.number_of_sites, resolution])
self.p_TiN_given_G = np.zeros([self.number_of_sites, resolution])
self.p_TiN_given_het = np.zeros([self.number_of_sites, resolution])
self.p_artifact = np.zeros([self.number_of_sites, 1])
# likelihood
self.TiN_likelihood = np.zeros([resolution, 1])
