def smn_cn_caller(
bam, region_dic, gmm_parameter,
snp_db, variant_db, threads, count_file=None, reference_fasta=None):
"""Return SMN CN calls for each sample."""
# 1. read counting, normalization
if count_file is not None:
bamfile = open_alignment_file(bam, reference_fasta)
reads = bamfile.fetch()
read_length = get_read_length(reads)
bamfile.close()
normalized_depth = get_normed_depth_from_count(
count_file, region_dic, read_length, gc_correct=False)
else:
normalized_depth = get_normed_depth(
bam, region_dic, threads, reference=reference_fasta, gc_correct=False)
# 2. GMM and CN call
cn_call = namedtuple(
'cn_call', 'exon16_cn exon16_depth exon78_cn exon78_depth'
)
gmm_exon16 = Gmm()
gmm_exon16.set_gmm_par(gmm_parameter, 'exon1-6')
gcall_exon16 = gmm_exon16.gmm_call(normalized_depth.normalized['exon16'])
gmm_exon78 = Gmm()
gmm_exon78.set_gmm_par(gmm_parameter, 'exon7-8')
gcall_exon78 = gmm_exon78.gmm_call(normalized_depth.normalized['exon78'])
raw_cn_call = cn_call(
gcall_exon16.cn, gcall_exon16.depth_value,
gcall_exon78.cn, gcall_exon78.depth_value
)
# 3. Get SNP ratios
smn1_read_count, smn2_read_count = get_supporting_reads(
bam, snp_db.dsnp1, snp_db.dsnp2, snp_db.nchr, snp_db.dindex, reference=reference_fasta
)
smn1_fraction = get_fraction(smn1_read_count, smn2_read_count)
var_ref_count, var_alt_count = get_supporting_reads(
bam, variant_db.dsnp1, variant_db.dsnp2, variant_db.nchr,
variant_db.dindex, reference=reference_fasta
)
# 4. Call CN of SMN1 and SMN2
final_call = get_smn12_call(
raw_cn_call, smn1_read_count, smn2_read_count,
var_ref_count, var_alt_count,
normalized_depth.mediandepth
)
# 5. Prepare final call set
sample_call = namedtuple(
'sample_call',
'Coverage_MAD \
Full_length_CN_raw Total_CN_raw \
SMN1_read_support SMN2_read_support SMN1_fraction \
g27134TG_REF_count g27134TG_ALT_count'
)
sample_cn_call = sample_call(
round(normalized_depth.mad, 3),
raw_cn_call.exon78_depth, raw_cn_call.exon16_depth,
smn1_read_count, smn2_read_count, [round(a, 2) for a in smn1_fraction],
var_ref_count, var_alt_count
)
doutput = sample_cn_call._asdict()
doutput.update(final_call._asdict())
return doutput
def d6_star_caller(bam,
call_parameters,
threads,
count_file=None,
reference_fasta=None):
"""Return CYP2D6 star allele diplotype calls for each sample."""
d6_call = namedtuple(
"d6_call",
"Coverage_MAD Median_depth Total_CN Spacer_CN Total_CN_raw \
Spacer_CN_raw Variants_called CNV_group Genotype Filter Raw_star_allele \
Call_info Exon9_CN CNV_consensus d67_snp_call d67_snp_raw \
Variant_raw_count",
)
# 1. Read counting and normalization
bamfile = open_alignment_file(bam, reference_fasta)
if count_file is not None:
reads = bamfile.fetch()
read_length = get_read_length(reads)
normalized_depth = get_normed_depth_from_count(
count_file, call_parameters.region_dic, read_length)
else:
normalized_depth = get_normed_depth(bam,
call_parameters.region_dic,
threads,
reference=reference_fasta)
# no-call after normalizaton
if normalized_depth.normalized["d67"] is None:
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
)
return sample_call
# 2. GMM and CN call
# There are two regions to call CN based on depth: total CYP2D6+CYP2D7, and CYP2D7 spacer region
cn_call = namedtuple("cn_call", "d67_cn d67_depth spacer_cn spacer_depth")
gmm_d67 = Gmm()
gmm_d67.set_gmm_par(call_parameters.gmm_parameter, "d67")
gcall_d67 = gmm_d67.gmm_call(normalized_depth.normalized["d67"])
gmm_spacer = Gmm()
gmm_spacer.set_gmm_par(call_parameters.gmm_parameter, "spacer")
gcall_spacer = gmm_spacer.gmm_call(normalized_depth.normalized["spacer"])
high_cn_low_confidence = False
if gcall_d67.cn is None and gcall_d67.depth_value > HIGH_CN_DEPTH_THRESHOLD:
high_cn_low_confidence = True
raw_cn_call = cn_call(
int(round(gcall_d67.depth_value)),
gcall_d67.depth_value,
gcall_spacer.cn,
gcall_spacer.depth_value,
)
else:
raw_cn_call = cn_call(
gcall_d67.cn,
gcall_d67.depth_value,
gcall_spacer.cn,
gcall_spacer.depth_value,
)
# 3. Get allele counts at D6/D7 SNP (base difference) sites and target variant sites
# D6/D7 base difference sites. Get read counts at both D6/D7 positions.
snp_db = call_parameters.snp_db
snp_d6, snp_d7 = get_supporting_reads(
bam,
snp_db.dsnp1,
snp_db.dsnp2,
snp_db.nchr,
snp_db.dindex,
reference=reference_fasta,
)
site42126938 = [snp_d6[VAR42126938_SITE], snp_d7[VAR42126938_SITE]]
snp_d6.pop(VAR42126938_SITE)
snp_d6.pop(VAR42126938_SITE - 1)
snp_d7.pop(VAR42126938_SITE)
snp_d7.pop(VAR42126938_SITE - 1)
# Variants not in homology regions. Get read counts only at D6 positions.
var_db = call_parameters.var_db
var_alt, var_ref = get_supporting_reads_single_region(
bam,
var_db.dsnp1,
var_db.nchr,
var_db.dindex,
reference=reference_fasta)
# Look more carefully for insertions at 42128936 from reads
var_list = call_parameters.var_list
ref_read, long_ins_read, short_ins_read = get_allele_counts_42128936(
bamfile, call_parameters.genome)
if "g.42128936-42128937insGGGGCGAAAGGGGCGAAA" in var_list:
long_ins_index = var_list.index(
"g.42128936-42128937insGGGGCGAAAGGGGCGAAA")
var_alt[long_ins_index] = long_ins_read
var_ref[long_ins_index] = short_ins_read + ref_read
if "g.42128936-42128937insGGGGCGAAA" in var_list:
short_ins_index = var_list.index("g.42128936-42128937insGGGGCGAAA")
var_alt[short_ins_index] = short_ins_read
var_ref[short_ins_index] = long_ins_read + ref_read
# Variants in homology regions. Get read counts at both D6/D7 positions.
var_homo_db = call_parameters.var_homo_db
var_homo_alt, var_homo_ref = get_supporting_reads(
bam,
var_homo_db.dsnp1,
var_homo_db.dsnp2,
var_homo_db.nchr,
var_homo_db.dindex,
reference=reference_fasta,
)
# This ordered dictionary is for final reporting.
raw_count = OrderedDict()
for i in range(len(call_parameters.var_list)):
if i < len(var_alt):
raw_count.setdefault(var_list[i],
"%i,%i" % (var_alt[i], var_ref[i]))
else:
raw_count.setdefault(
var_list[i],
"%i,%i" % (var_homo_alt[i - len(var_alt)],
var_homo_ref[i - len(var_alt)]),
)
raw_count.setdefault("g.42126938C>T",
"%i,%i" % (site42126938[0], site42126938[1]))
# no-call due to total copy number calling
if raw_cn_call.d67_cn is None:
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
raw_cn_call.d67_cn,
raw_cn_call.spacer_cn,
raw_cn_call.d67_depth,
raw_cn_call.spacer_depth,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
raw_count,
)
return sample_call
# 4. Call CNV and hybrids
d6_fraction = get_fraction(snp_d6, snp_d7)
raw_d6_cn = [round(raw_cn_call.d67_cn * a, 3) for a in d6_fraction]
cn_call_snp = call_cn_snp(raw_cn_call.d67_cn, snp_d6, snp_d7)
# exon9gc
exon9gc_call_stringent = call_exon9gc(snp_d6[EXON9_SITE1],
snp_d7[EXON9_SITE1],
raw_cn_call.d67_cn)
cnvtag, consensus = get_cnvtag(
raw_cn_call.d67_cn,
raw_d6_cn,
cn_call_snp,
exon9gc_call_stringent,
raw_cn_call.spacer_cn,
)
# no-call due to CNV group calling
if cnvtag is None or cnvtag not in CNV_ACCEPTED:
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
raw_cn_call.d67_cn,
raw_cn_call.spacer_cn,
raw_cn_call.d67_depth,
raw_cn_call.spacer_depth,
None,
cnvtag,
None,
None,
None,
None,
exon9gc_call_stringent,
",".join(str(a) for a in consensus),
",".join(str(a) for a in cn_call_snp),
",".join(str(a) for a in raw_d6_cn),
raw_count,
)
return sample_call
# 5. Call variants
# homology region
cn_call_var_homo = call_cn_var_homo(raw_cn_call.d67_cn, var_homo_alt,
var_homo_ref)
# non-homology region
cn_call_var = call_cn_var(cnvtag, var_alt, var_ref, var_list, var_db)
# call g.42126938C>T
if cnvtag in ["star5", "cn2"]:
var42126938, G_haplotype = call_var42126938(
bamfile,
cnvtag,
site42126938,
snp_db,
[VAR42126938_SITE - 2, VAR42126938_SITE - 1, VAR42126938_SITE],
)
else:
var42126938 = []
G_haplotype = False
# 6. Call star allele
total_callset = get_called_variants(var_list, cn_call_var)
called_var_homo = get_called_variants(var_list, cn_call_var_homo,
len(cn_call_var))
total_callset += called_var_homo
total_callset += var42126938
exon9_values = namedtuple(
"exon9_values",
"exon9_cn exon9cn_in_consensus exon9_raw_site1 exon9_raw_site2")
star_called = match_star(
total_callset,
cnvtag,
raw_cn_call.spacer_cn,
call_parameters.star_combinations,
exon9_values(
exon9gc_call_stringent,
consensus.exon9_and_downstream,
raw_d6_cn[EXON9_SITE1],
raw_d6_cn[EXON9_SITE2],
),
)
genotype_filter = None
# no-call due to star allele matching
if "no_match" in star_called[
0]: # or star_called[0] == 'more_than_one_match':
final_star_allele_call = None
elif (star_called[0] == "more_than_one_match"
and star_called[-1] == "*1/*32;*27/*41"):
genotype_filter = "PASS"
if G_haplotype:
# Variants are on the sample haplotype
final_star_allele_call = "*1/*32"
else:
final_star_allele_call = "*27/*41"
else:
final_star_allele_call = star_called[-1]
if ";" in final_star_allele_call:
genotype_filter = "More_than_one_possible_genotype"
elif "/" not in final_star_allele_call:
genotype_filter = "Not_assigned_to_haplotypes"
elif high_cn_low_confidence:
genotype_filter = "LowQ_high_CN"
else:
genotype_filter = "PASS"
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
raw_cn_call.d67_cn,
raw_cn_call.spacer_cn,
raw_cn_call.d67_depth,
raw_cn_call.spacer_depth,
star_called.variants_called.split(),
cnvtag,
final_star_allele_call,
genotype_filter,
star_called.raw_call,
star_called.call_info,
exon9gc_call_stringent,
",".join(str(a) for a in consensus),
",".join(str(a) for a in cn_call_snp),
",".join(str(a) for a in raw_d6_cn),
raw_count,
)
bamfile.close()
return sample_call
def d6_star_caller(
bam, call_parameters, threads, count_file=None, reference_fasta=None, index_name=None
):
"""Return CYP2D6 star allele diplotype calls for each sample."""
d6_call = namedtuple(
"d6_call",
"Coverage_MAD Median_depth Total_CN Spacer_CN Total_CN_raw \
Spacer_CN_raw Variants_called CNV_group Genotype Filter Raw_star_allele \
Call_info Exon9_CN CNV_consensus d67_snp_call d67_snp_raw \
Variant_raw_count",
)
# 1. Read counting and normalization
bamfile = open_alignment_file(bam, reference_fasta, index_filename=index_name)
if count_file is not None:
reads = bamfile.fetch()
read_length = get_read_length(reads)
normalized_depth = get_normed_depth_from_count(
count_file, call_parameters.region_dic, read_length
)
else:
normalized_depth = get_normed_depth(
bam, call_parameters.region_dic, threads, reference=reference_fasta
)
# no-call after normalizaton
if normalized_depth.normalized["d67"] is None:
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
)
return sample_call
# 2. GMM and CN call
# There are two regions to call CN based on depth: total CYP2D6+CYP2D7, and CYP2D7 spacer region
cn_call = namedtuple("cn_call", "d67_cn d67_depth spacer_cn spacer_depth")
gmm_d67 = Gmm()
gmm_d67.set_gmm_par(call_parameters.gmm_parameter, "d67")
gcall_d67 = gmm_d67.gmm_call(normalized_depth.normalized["d67"])
gmm_spacer = Gmm()
gmm_spacer.set_gmm_par(call_parameters.gmm_parameter, "spacer")
gcall_spacer = gmm_spacer.gmm_call(normalized_depth.normalized["spacer"])
high_cn_low_confidence = False
if gcall_d67.cn is None and gcall_d67.depth_value > HIGH_CN_DEPTH_THRESHOLD:
high_cn_low_confidence = True
raw_cn_call = cn_call(
int(round(gcall_d67.depth_value)),
gcall_d67.depth_value,
gcall_spacer.cn,
gcall_spacer.depth_value,
)
else:
raw_cn_call = cn_call(
gcall_d67.cn,
gcall_d67.depth_value,
gcall_spacer.cn,
gcall_spacer.depth_value,
)
# 3. Get allele counts at D6/D7 SNP (base difference) sites and target variant sites
# D6/D7 base difference sites. Get read counts at both D6/D7 positions.
snp_db = call_parameters.snp_db
snp_d6, snp_d7 = get_supporting_reads(
bamfile, snp_db.dsnp1, snp_db.dsnp2, snp_db.nchr, snp_db.dindex
)
# Variants not in homology regions. Get read counts only at D6 positions.
var_db = call_parameters.var_db
var_alt, var_ref, var_alt_forward, var_alt_reverse = get_supporting_reads_single_region(
bamfile, var_db.dsnp1, var_db.nchr, var_db.dindex
)
# Look more carefully for insertions at 42128936 from reads
var_list = call_parameters.var_list
ref_read, long_ins_read, short_ins_read = get_allele_counts_var42128936(
bamfile, call_parameters.genome
)
var_alt, var_ref = update_var42128936(
var_list, var_alt, var_ref, ref_read, long_ins_read, short_ins_read
)
# Variants in homology regions. Get read counts at both D6/D7 positions.
var_homo_db = call_parameters.var_homo_db
var_homo_alt, var_homo_ref = get_supporting_reads(
bamfile,
var_homo_db.dsnp1,
var_homo_db.dsnp2,
var_homo_db.nchr,
var_homo_db.dindex,
)
# This ordered dictionary is for final reporting.
raw_count = OrderedDict()
non_homology_variant_count = len(var_alt)
for i in range(len(call_parameters.var_list)):
if i < non_homology_variant_count:
if var_list[i] in NOISY_VAR:
raw_count.setdefault(
var_list[i],
"%i(%i:%i),%i"
% (var_alt[i], var_alt_forward[i], var_alt_reverse[i], var_ref[i]),
)
else:
raw_count.setdefault(var_list[i], "%i,%i" % (var_alt[i], var_ref[i]))
else:
raw_count.setdefault(
var_list[i],
"%i,%i"
% (
var_homo_alt[i - non_homology_variant_count],
var_homo_ref[i - non_homology_variant_count],
),
)
# no-call due to total copy number calling
if raw_cn_call.d67_cn is None:
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
raw_cn_call.d67_cn,
raw_cn_call.spacer_cn,
raw_cn_call.d67_depth,
raw_cn_call.spacer_depth,
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
raw_count,
)
return sample_call
# 4. Call CNV and hybrids
d6_fraction = get_fraction(snp_d6, snp_d7)
raw_d6_cn = [round(raw_cn_call.d67_cn * a, 3) for a in d6_fraction]
cn_call_snp = call_cn_snp(raw_cn_call.d67_cn, snp_d6, snp_d7)
# exon9gc
exon9gc_call_stringent = call_exon9gc(
snp_d6[EXON9_SITE1 : EXON9_SITE2 + 1],
snp_d7[EXON9_SITE1 : EXON9_SITE2 + 1],
raw_cn_call.d67_cn,
)
cnvtag, consensus = get_cnvtag(
raw_cn_call.d67_cn,
raw_d6_cn,
cn_call_snp,
exon9gc_call_stringent,
raw_cn_call.spacer_cn,
)
# no-call due to CNV group calling
if cnvtag is None or cnvtag not in CNV_ACCEPTED:
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
raw_cn_call.d67_cn,
raw_cn_call.spacer_cn,
raw_cn_call.d67_depth,
raw_cn_call.spacer_depth,
None,
cnvtag,
None,
None,
None,
None,
exon9gc_call_stringent,
",".join(str(a) for a in consensus),
",".join(str(a) for a in cn_call_snp),
",".join(str(a) for a in raw_d6_cn),
raw_count,
)
return sample_call
# 5. Call variants
# homology region
cn_call_var_homo = call_cn_var_homo(raw_cn_call.d67_cn, var_homo_alt, var_homo_ref)
# non-homology region
cn_call_var = call_cn_var(
cnvtag, var_alt, var_ref, var_alt_forward, var_alt_reverse, var_list, var_db
)
# call haplotypes
haplotype_db = call_parameters.haplotype_db
site42126938_count, var42126938, var42126938_G_haplotype = call_var42126938(
bamfile, raw_cn_call.d67_cn, haplotype_db["g.42126938C>T"]
)
raw_count.setdefault(
"g.42126938C>T", "%i,%i" % (site42126938_count[1], site42126938_count[0])
)
site42127526_count, site42127556_count, var42127526 = call_var42127526_var42127556(
bamfile, cnvtag, haplotype_db["g.42127526C>T_g.42127556T>C"]
)
raw_count.setdefault(
"g.42127526C>T", "%i,%i" % (site42127526_count[1], site42127526_count[0])
)
raw_count.setdefault(
"g.42127556T>C", "%i,%i" % (site42127556_count[1], site42127556_count[0])
)
var42127803_diff_haplotype = call_var42127803hap(
bamfile, cnvtag, haplotype_db["g.42127803C>T"]
)
# 6. Call star allele
total_callset = get_called_variants(var_list, cn_call_var)
called_var_homo = get_called_variants(var_list, cn_call_var_homo, len(cn_call_var))
total_callset += called_var_homo
total_callset += var42126938
total_callset += var42127526
star_called = match_star(
total_callset,
cnvtag,
raw_cn_call.spacer_cn,
call_parameters.star_combinations,
exon9_values(
exon9gc_call_stringent,
consensus.exon9_and_downstream,
raw_d6_cn[EXON9_SITE1],
raw_d6_cn[EXON9_SITE2],
),
var42126938_G_haplotype,
var42127803_diff_haplotype,
)
genotype_filter = None
# no-call due to star allele matching
if "no_match" in star_called[0]: # or star_called[0] == 'more_than_one_match':
final_star_allele_call = None
else:
final_star_allele_call = star_called[-1]
if ";" in final_star_allele_call:
genotype_filter = "More_than_one_possible_genotype"
elif "/" not in final_star_allele_call:
genotype_filter = "Not_assigned_to_haplotypes"
elif high_cn_low_confidence:
genotype_filter = "LowQ_high_CN"
else:
genotype_filter = "PASS"
sample_call = d6_call(
normalized_depth.mad,
normalized_depth.mediandepth,
raw_cn_call.d67_cn,
raw_cn_call.spacer_cn,
raw_cn_call.d67_depth,
raw_cn_call.spacer_depth,
star_called.variants_called.split(),
cnvtag,
final_star_allele_call,
genotype_filter,
star_called.raw_call,
star_called.call_info,
exon9gc_call_stringent,
",".join(str(a) for a in consensus),
",".join(str(a) for a in cn_call_snp),
",".join(str(a) for a in raw_d6_cn),
raw_count,
)
bamfile.close()
return sample_call