def add_snp(snp, position, sequence):
if snp.alt != ".":
genotype = "".join(vcfwrap.get_ml_genotype(snp))
encoded = sequences.encode_genotype(genotype)
else:
encoded = snp.ref
sequence[position] = encoded
def add_snp(snp, position, sequence):
if snp.alt != ".":
genotype = "".join(vcfwrap.get_ml_genotype(snp))
encoded = sequences.encode_genotype(genotype)
else:
encoded = snp.ref
sequence[position] = encoded
def main(argv):
if len(argv) != 2:
sys.stderr.write("Usage: %s <BED-file> <VCF.bgz>\n")
return 1
sites = 0
sites_non_ref = 0
sites_homo_non_ref = 0
sites_het_one_non_ref = 0
sites_het_two_non_ref = 0
vcf_records = pysam.Tabixfile(argv[1])
bed_records = read_bed_records(argv[0])
for record in select_vcf_records(bed_records, vcf_records):
if record.alt != '.':
# Get the most likely diploid genotype
nt_a, nt_b = get_ml_genotype(record)
if (nt_a, nt_b) == ('N', 'N'):
# Skip sites with no most likely genotype
continue
sites += 1
sites_non_ref += 1
if nt_a == nt_b:
sites_homo_non_ref += 1
elif record.ref not in (nt_a, nt_b):
sites_het_two_non_ref += 1
else:
sites_het_one_non_ref += 1
else:
# Heterozygous for the reference allele
sites += 1
print
print "%i sites kept after filtering:" % (sites, )
print " % 10i homozygous sites containing the reference allele (%.2f%%)" % (
sites - sites_non_ref, 100.0 * (sites - sites_non_ref) / float(sites))
print " % 10i heterozygous sites containing the reference and a non-reference allele (%.2f%%)" % (
sites_het_one_non_ref, (100.0 * sites_het_one_non_ref) / sites)
print " % 10i homozygous sites containing a single non-reference allele (%.2f%%)" % (
sites_homo_non_ref, (100.0 * sites_homo_non_ref) / sites)
print " % 10i heterozygous sites containing two different non-reference alleles (%.2f%%)" % (
sites_het_two_non_ref, (100.0 * sites_het_two_non_ref) / sites)
def add_indel(options, bed, indel, sequence):
if indel.alt == ".":
return
genotype = vcfwrap.get_ml_genotype(indel)
if genotype[0] != genotype[1]:
# No way to represent heterozygous indels
return
elif genotype[0] == "N":
# No most likely genotype
return
# Note that bed.end is a past-the-end coordinate
start = max(0, bed.start - options.padding)
# FIXME: parse_indel only supports a single 'alt' values
indel.alt = genotype[0]
indel = vcfwrap.parse_indel(indel)
if indel.in_reference:
del_start = max(indel.pos + 1, bed.start)
del_end = min(indel.pos + 1 + len(indel.what), bed.end)
if del_start >= del_end:
# Deletion does not cover any bases of interest
return
elif options.whole_codon_indels_only:
if (del_end - del_start) % 3:
# Non-codon sized overlap with area of interest
return
for position in range(del_start, del_end):
sequence[position - start] = ""
elif (len(indel.what) % 3 == 0) or not options.whole_codon_indels_only:
# parse_indel assumes that the insertion is always the first possible
# base when multiple positions are possible. As a consequence, the
# position may be before start, with the rest of the bases overlapping
# the current sequence. For example:
# ref = ATTT
# alt = ATTTT
# It is assumed that the insertion (_) happened thus:
# interpretation = A_TTT
if indel.pos >= start:
sequence[indel.pos - start] += indel.what
def add_indel(options, bed, indel, sequence):
if indel.alt == ".":
return
genotype = vcfwrap.get_ml_genotype(indel)
if genotype[0] != genotype[1]:
# No way to represent heterozygous indels
return
elif genotype[0] == "N":
# No most likely genotype
return
# Note that bed.end is a past-the-end coordinate
start = max(0, bed.start - options.padding)
# FIXME: parse_indel only supports a single 'alt' values
indel.alt = genotype[0]
indel = vcfwrap.parse_indel(indel)
if indel.in_reference:
del_start = max(indel.pos + 1, bed.start)
del_end = min(indel.pos + 1 + len(indel.what), bed.end)
if del_start >= del_end:
# Deletion does not cover any bases of interest
return
elif options.whole_codon_indels_only:
if (del_end - del_start) % 3:
# Non-codon sized overlap with area of interest
return
for position in range(del_start, del_end):
sequence[position - start] = ""
elif (len(indel.what) % 3 == 0) or not options.whole_codon_indels_only:
# parse_indel assumes that the insertion is always the first possible
# base when multiple positions are possible. As a consequence, the
# position may be before start, with the rest of the bases overlapping
# the current sequence. For example:
# ref = ATTT
# alt = ATTTT
# It is assumed that the insertion (_) happened thus:
# interpretation = A_TTT
if indel.pos >= start:
sequence[indel.pos - start] += indel.what
def main(argv):
if len(argv) != 2:
sys.stderr.write("Usage: %s <BED-file> <VCF.bgz>\n")
return 1
sites = 0
sites_non_ref = 0
sites_homo_non_ref = 0
sites_het_one_non_ref = 0
sites_het_two_non_ref = 0
vcf_records = pysam.Tabixfile(argv[1])
bed_records = read_bed_records(argv[0])
for record in select_vcf_records(bed_records, vcf_records):
if record.alt != '.':
# Get the most likely diploid genotype
nt_a, nt_b = get_ml_genotype(record)
if (nt_a, nt_b) == ('N', 'N'):
# Skip sites with no most likely genotype
continue
sites += 1
sites_non_ref += 1
if nt_a == nt_b:
sites_homo_non_ref += 1
elif record.ref not in (nt_a, nt_b):
sites_het_two_non_ref += 1
else:
sites_het_one_non_ref += 1
else:
# Heterozygous for the reference allele
sites += 1
print
print "%i sites kept after filtering:" % (sites,)
print " % 10i homozygous sites containing the reference allele (%.2f%%)" % (sites - sites_non_ref, 100.0 * (sites - sites_non_ref) / float(sites))
print " % 10i heterozygous sites containing the reference and a non-reference allele (%.2f%%)" % (sites_het_one_non_ref, (100.0 * sites_het_one_non_ref) / sites)
print " % 10i homozygous sites containing a single non-reference allele (%.2f%%)" % (sites_homo_non_ref, (100.0 * sites_homo_non_ref) / sites)
print " % 10i heterozygous sites containing two different non-reference alleles (%.2f%%)" % (sites_het_two_non_ref, (100.0 * sites_het_two_non_ref) / sites)
def _filter_by_properties(options, vcfs, frequencies):
"""Filters a list of SNPs/indels based on the various properties recorded in
the info column, and others. This mirrors most of the filtering carried out
by vcfutils.pl varFilter."""
for vcf in vcfs:
if float(vcf.qual) < options.min_quality:
_mark_as_filtered(vcf, "q:%i" % options.min_quality)
properties = {}
for field in vcf.info.split(";"):
if "=" in field:
key, value = field.split("=")
else:
key, value = field, None
properties[key] = value
read_depth = float(properties["DP"])
if options.min_read_depth > read_depth:
_mark_as_filtered(vcf, "d:%i" % options.min_read_depth)
elif options.max_read_depth < read_depth:
_mark_as_filtered(vcf, "D:%i" % options.max_read_depth)
if "MQ" in properties:
if float(properties["MQ"]) < options.min_mapping_quality:
_mark_as_filtered(vcf, "Q:%i" % options.min_mapping_quality)
if "PV4" in properties:
pv4 = [float(value) for value in properties["PV4"].split(",")]
if (pv4[0] < options.min_strand_bias):
_mark_as_filtered(vcf, "1:%e" % options.min_strand_bias)
if (pv4[1] < options.min_baseq_bias):
_mark_as_filtered(vcf, "2:%e" % options.min_baseq_bias)
if (pv4[2] < options.min_mapq_bias):
_mark_as_filtered(vcf, "3:%e" % options.min_mapq_bias)
if (pv4[3] < options.min_end_distance_bias):
_mark_as_filtered(vcf, "4:%e" % options.min_end_distance_bias)
if vcf.alt != ".":
ref_fw, ref_rev, alt_fw, alt_rev = map(int, properties["DP4"].split(","))
if (alt_fw + alt_rev) < options.min_num_alt_bases:
_mark_as_filtered(vcf, "a:%i" % options.min_num_alt_bases)
ml_genotype = vcfwrap.get_ml_genotype(vcf)
if (ml_genotype == ("N", "N")) and not options.keep_ambigious_genotypes:
# No most likely genotype
_mark_as_filtered(vcf, "k")
if (ml_genotype[0] != ml_genotype[1]):
if vcf.contig in options.homozygous_chromosome:
_mark_as_filtered(vcf, "HET")
# Filter by frequency of minor allele
if vcf.ref in ml_genotype:
n_minor = min(ref_fw + ref_rev, alt_fw + alt_rev)
n_major = max(ref_fw + ref_rev, alt_fw + alt_rev)
if (n_minor / float(n_minor + n_major)) < options.min_allele_frequency:
_mark_as_filtered(vcf, "f:%.4f" % options.min_allele_frequency)
else:
state = frequencies.frequency_is_valid(vcf.contig, vcf.pos, vcf.ref, *ml_genotype)
if state is frequencies.INVALID:
_mark_as_filtered(vcf, "f:%.4f" % options.min_allele_frequency)
elif state is frequencies.NA:
if _mark_as_filtered(vcf, "F:%.4f" % options.min_allele_frequency):
sys.stderr.write("WARNING: Could not determine allele-counts for SNP at %s:%s, filtering ...\n" % (vcf.contig, vcf.pos + 1))
def _filter_by_properties(options, vcfs, frequencies):
"""Filters a list of SNPs/indels based on the various properties recorded in
the info column, and others. This mirrors most of the filtering carried out
by vcfutils.pl varFilter."""
for vcf in vcfs:
if float(vcf.qual) < options.min_quality:
_mark_as_filtered(vcf, "q:%i" % options.min_quality)
properties = {}
for field in vcf.info.split(";"):
if "=" in field:
key, value = field.split("=")
else:
key, value = field, None
properties[key] = value
read_depth = float(properties["DP"])
if options.min_read_depth > read_depth:
_mark_as_filtered(vcf, "d:%i" % options.min_read_depth)
elif options.max_read_depth < read_depth:
_mark_as_filtered(vcf, "D:%i" % options.max_read_depth)
if "MQ" in properties:
if float(properties["MQ"]) < options.min_mapping_quality:
_mark_as_filtered(vcf, "Q:%i" % options.min_mapping_quality)
if "PV4" in properties:
pv4 = [float(value) for value in properties["PV4"].split(",")]
if (pv4[0] < options.min_strand_bias):
_mark_as_filtered(vcf, "1:%e" % options.min_strand_bias)
if (pv4[1] < options.min_baseq_bias):
_mark_as_filtered(vcf, "2:%e" % options.min_baseq_bias)
if (pv4[2] < options.min_mapq_bias):
_mark_as_filtered(vcf, "3:%e" % options.min_mapq_bias)
if (pv4[3] < options.min_end_distance_bias):
_mark_as_filtered(vcf, "4:%e" % options.min_end_distance_bias)
if vcf.alt != ".":
ref_fw, ref_rev, alt_fw, alt_rev = map(
int, properties["DP4"].split(","))
if (alt_fw + alt_rev) < options.min_num_alt_bases:
_mark_as_filtered(vcf, "a:%i" % options.min_num_alt_bases)
ml_genotype = vcfwrap.get_ml_genotype(vcf)
if (ml_genotype
== ("N", "N")) and not options.keep_ambigious_genotypes:
# No most likely genotype
_mark_as_filtered(vcf, "k")
if (ml_genotype[0] != ml_genotype[1]):
if vcf.contig in options.homozygous_chromosome:
_mark_as_filtered(vcf, "HET")
# Filter by frequency of minor allele
if vcf.ref in ml_genotype:
n_minor = min(ref_fw + ref_rev, alt_fw + alt_rev)
n_major = max(ref_fw + ref_rev, alt_fw + alt_rev)
if (n_minor / float(n_minor + n_major)
) < options.min_allele_frequency:
_mark_as_filtered(
vcf, "f:%.4f" % options.min_allele_frequency)
else:
state = frequencies.frequency_is_valid(
vcf.contig, vcf.pos, vcf.ref, *ml_genotype)
if state is frequencies.INVALID:
_mark_as_filtered(
vcf, "f:%.4f" % options.min_allele_frequency)
elif state is frequencies.NA:
if _mark_as_filtered(
vcf, "F:%.4f" % options.min_allele_frequency):
sys.stderr.write(
"WARNING: Could not determine allele-counts for SNP at %s:%s, filtering ...\n"
% (vcf.contig, vcf.pos + 1))
