def prune_allrefs(vcf, fout):
for record in vcf.fetch():
if is_biallelic(record):
if record.info['AC'][0] > 0:
fout.write(record)
else:
if record.info['CN_NONREF_FREQ'] > 0:
fout.write(record)
def drop_nonref_gts(vcf, fout):
NULL_GT = [(0, 0), (None, None), (0, ), (None, ), (None, 2)]
samples = [s for s in vcf.header.samples]
# for record in vcf.fetch():
# nonref = svu.get_called_samples(record)
# if len(nonref) > 0:
# fout.write(record)
for record in vcf.fetch():
for s in samples:
if is_biallelic(record):
if record.samples[s]['GT'] not in NULL_GT:
fout.write(record)
break
else:
if record.samples[s]['CN'] != 2:
fout.write(record)
break
def calc_allele_freq(record, samples, prefix=None, hemi=False):
"""
Computes allele frequencies for a single record based on a list of samples
"""
# Treat biallelic and multiallelic records differently
# For biallelic sites, count number of non-ref, non-no-call GTs
if svu.is_biallelic(record):
# Get all sample GTs
GTs = [record.samples[s]['GT'] for s in samples]
# Count alleles & genotypes
AC = 0
AN = 0
n_alt_count_0 = 0
n_alt_count_1 = 0
n_alt_count_2 = 0
n_gts_with_gt_0_alts = 0 # Used specifically for hemizygous sites
for GT in GTs:
AN += len([
allele for allele in GT if allele != '.' and allele is not None
])
AC += len([
allele for allele in GT
if allele != '.' and allele != 0 and allele is not None
])
if GT == (0, 0):
n_alt_count_0 += 1
if len([allele for allele in GT if allele == 0 and allele != '.' and allele is not None]) == 1 \
and len([allele for allele in GT if allele != 0 and allele != '.' and allele is not None]) == 1:
n_alt_count_1 += 1
n_gts_with_gt_0_alts += 1
if len([
allele for allele in GT
if allele != 0 and allele != '.' and allele is not None
]) == 2:
n_alt_count_2 += 1
n_gts_with_gt_0_alts += 1
# Adjust hemizygous allele number and allele count, if optioned
if hemi:
AN = AN / 2
# For hemizygous sites, AC must be the sum of all non-reference *genotypes*, not alleles
AC = n_gts_with_gt_0_alts
# Calculate allele frequency
if AN > 0:
AF = AC / AN
AF = round(AF, 6)
else:
AF = 0
# Add AN, AC, and AF to INFO field
record.info[(prefix + '_' if prefix else '') + 'AN'] = AN
record.info[(prefix + '_' if prefix else '') + 'AC'] = AC
record.info[(prefix + '_' if prefix else '') + 'AF'] = AF
# Calculate genotype frequencies
n_bi_genos = n_alt_count_0 + n_alt_count_1 + n_alt_count_2
if n_bi_genos > 0:
freq_homref = n_alt_count_0 / n_bi_genos
freq_het = n_alt_count_1 / n_bi_genos
freq_homalt = n_alt_count_2 / n_bi_genos
else:
freq_homref = 0
freq_het = 0
freq_homalt = 0
if hemi:
freq_hemialt = freq_het + freq_homalt
# Add N_BI_GENOS, N_HOMREF, N_HET, N_HOMALT, FREQ_HOMREF, FREQ_HET, and FREQ_HOMALT to INFO field
record.info[(prefix + '_' if prefix else '') +
'N_BI_GENOS'] = n_bi_genos
if hemi:
record.info[(prefix + '_' if prefix else '') +
'N_HEMIREF'] = n_alt_count_0
record.info[(prefix + '_' if prefix else '') +
'N_HEMIALT'] = n_gts_with_gt_0_alts
record.info[(prefix + '_' if prefix else '') +
'FREQ_HEMIREF'] = freq_homref
record.info[(prefix + '_' if prefix else '') +
'FREQ_HEMIALT'] = freq_hemialt
record.info[(prefix + '_' if prefix else '') +
'N_HOMREF'] = n_alt_count_0
record.info[(prefix + '_' if prefix else '') + 'N_HET'] = n_alt_count_1
record.info[(prefix + '_' if prefix else '') +
'N_HOMALT'] = n_alt_count_2
record.info[(prefix + '_' if prefix else '') +
'FREQ_HOMREF'] = freq_homref
record.info[(prefix + '_' if prefix else '') + 'FREQ_HET'] = freq_het
record.info[(prefix + '_' if prefix else '') +
'FREQ_HOMALT'] = freq_homalt
# Multiallelic sites should reference FORMAT:CN rather than GT
# Compute CN_NUMBER, CN_NONREF_COUNT, CN_NONREF_FREQ, and CN_COUNT/CN_FREQ for each copy state
else:
# Get all sample CNs and remove Nones
CNs_wNones = [record.samples[s]['CN'] for s in samples]
CNs = [
c for c in CNs_wNones if c is not None and c not in '. NA'.split()
]
if len(CNs) == 0:
nonnull_CNs, nonref_CN_count, nonref_CN_freq = [0] * 3
CN_dist = (0, )
CN_freqs = (0, )
else:
# Count number of samples per CN and total CNs observed
CN_counts = dict(Counter(CNs))
nonnull_CNs = len(CNs)
# Get max observed CN and enumerate counts/frequencies per CN as list starting from CN=0
max_CN = max([int(k) for k, v in CN_counts.items()])
CN_dist = [int(CN_counts.get(k, 0)) for k in range(max_CN + 1)]
CN_freqs = [round(v / nonnull_CNs, 6) for v in CN_dist]
# Get total non-reference CN counts and freq
if hemi:
ref_CN = 1
else:
ref_CN = 2
nonref_CN_count = sum([
int(CN_counts.get(k, 0)) for k in range(max_CN + 1)
if k != ref_CN
])
nonref_CN_freq = round(nonref_CN_count / nonnull_CNs, 6)
# Add values to INFO field
record.info[(prefix + '_' if prefix else '') +
'CN_NUMBER'] = nonnull_CNs
record.info[(prefix + '_' if prefix else '') +
'CN_COUNT'] = tuple(CN_dist)
record.info[(prefix + '_' if prefix else '') +
'CN_FREQ'] = tuple(CN_freqs)
record.info[(prefix + '_' if prefix else '') +
'CN_NONREF_COUNT'] = nonref_CN_count
record.info[(prefix + '_' if prefix else '') +
'CN_NONREF_FREQ'] = nonref_CN_freq
return record
def gather_allele_freqs(record,
samples,
males_set,
females_set,
parbt,
pop_dict,
pops,
sex_chroms,
no_combos=False):
"""
Wrapper to compute allele frequencies for all sex & population pairings
"""
# Add PAR annotation to record (if optioned)
if record.chrom in sex_chroms and len(parbt) > 0:
if in_par(record, parbt):
rec_in_par = True
record.info['PAR'] = True
else:
rec_in_par = False
else:
rec_in_par = False
# Get allele frequencies for all populations
calc_allele_freq(record, samples)
if len(males_set) > 0:
if record.chrom in sex_chroms and not rec_in_par:
calc_allele_freq(record, males_set, prefix='MALE', hemi=True)
else:
calc_allele_freq(record, males_set, prefix='MALE')
if len(females_set) > 0:
calc_allele_freq(record, females_set, prefix='FEMALE')
# Adjust global allele frequencies on sex chromosomes, if famfile provided
if record.chrom in sex_chroms and not rec_in_par \
and svu.is_biallelic(record) and len(males_set) + len(females_set) > 0:
update_sex_freqs(record)
# Get allele frequencies per population
if len(pops) > 0:
for pop in pops:
pop_samps = [s for s in samples if pop_dict.get(s, None) == pop]
calc_allele_freq(record, pop_samps, prefix=pop)
if len(males_set) > 0 and not no_combos:
if record.chrom in sex_chroms and not rec_in_par:
calc_allele_freq(
record,
list([s for s in pop_samps if s in males_set]),
prefix=pop + '_MALE',
hemi=True)
else:
calc_allele_freq(
record,
list([s for s in pop_samps if s in males_set]),
prefix=pop + '_MALE')
if len(females_set) > 0 and not no_combos:
calc_allele_freq(
record,
list([s for s in pop_samps if s in females_set]),
prefix=pop + '_FEMALE')
# Adjust per-pop allele frequencies on sex chromosomes, if famfile provided
if record.chrom in sex_chroms and not rec_in_par \
and svu.is_biallelic(record) and len(males_set) + len(females_set) > 0:
update_sex_freqs(record, pop=pop)
# Get POPMAX AF biallelic sites only
if svu.is_biallelic(record):
AFs = [record.info['{0}_AF'.format(pop)][0] for pop in pops]
popmax = max(AFs)
record.info['POPMAX_AF'] = popmax
return record
def cleanup_vcf(vcf,
fout,
callrates,
min_callrate_global=0.85,
min_callrate_smallDels=0.95):
# minus_samples = [s for s in vcf.header.samples if s not in plus_samples]
# male_minus_samples = [s for s in minus_samples if s not in male_samples]
for record in vcf:
#Move several filters from FILTER to INFO
for filt in filts_for_info:
if filt in record.filter:
record.info[filt] = True
#Move HIGH_SR_BACKGROUND
#Remove all HIGH_NOCALL_RATE and HIGH_SR_BACKGROUND tags from FILTER column
newfilts = [
filt for filt in record.filter if filt not in filts_to_remove
]
record.filter.clear()
for filt in newfilts:
record.filter.add(filt)
if len(record.filter) == 0:
record.filter.add('PASS')
# #Mark sites with low PCR+ call rate
# plus_callrate = get_call_rate(record, plus_samples)
# if plus_callrate < min_callrate:
# if 'LOW_PCRPLUS_CALL_RATE' not in record.info.keys():
# record.info.keys().append('LOW_PCRPLUS_CALL_RATE')
# record.info['LOW_PCRPLUS_CALL_RATE'] = True
#Mark sites with low PCR- call rate
if record.id in callrates.keys():
callrate = callrates[record.id]
#Mark small (300bp-1kb) deletions with stricter 5% null gt rate,
#and mark all other variants at specified null gt rate
if record.info['SVTYPE'] == 'DEL' \
and record.info['SVLEN'] < 1000 \
and record.info['SVLEN'] > 300:
if callrate < min_callrate_smallDels:
record.filter.add('LOW_CALL_RATE')
else:
if callrate < min_callrate_global:
record.filter.add('LOW_CALL_RATE')
#Recalibrate QUAL score for biallelic variants
if is_biallelic(record):
newQUAL = recal_qual_score(record)
if newQUAL is not None:
record.qual = newQUAL
#Only check for non-empty GTs for biallelic variants
if is_biallelic(record):
for s in record.samples:
if record.samples[s]['GT'] not in NULL_and_REF_GTs:
fout.write(record)
break
else:
fout.write(record)