def _prepare_variation(self, var):
"""private method to collect metrics for a single variant (var) in a VCF file.
Extracts variant information, variant impacts and extra fields for annotation.
"""
extra_fields = {}
# these metric require that genotypes are present in the file
call_rate = None
hwe_p_value = None
pi_hat = None
inbreeding_coeff = None
hom_ref = het = hom_alt = unknown = None
# only compute certain metrics if genoypes are available
if not self.args.no_genotypes and not self.args.no_load_genotypes:
hom_ref = var.num_hom_ref
hom_alt = var.num_hom_alt
het = var.num_het
unknown = var.num_unknown
call_rate = var.call_rate
aaf = var.aaf
hwe_p_value, inbreeding_coeff = \
popgen.get_hwe_likelihood(hom_ref, het, hom_alt, aaf)
pi_hat = var.nucl_diversity
else:
aaf = infotag.extract_aaf(var)
if not isinstance(aaf, (float, int)):
if aaf is not None:
aaf = max(aaf)
############################################################
# collect annotations from gemini's custom annotation files
# but only if the size of the variant is <= 50kb
############################################################
if var.end - var.POS < 50000:
pfam_domain = annotations.get_pfamA_domains(var)
cyto_band = annotations.get_cyto_info(var)
rs_ids = annotations.get_dbsnp_info(var)
clinvar_info = annotations.get_clinvar_info(var)
in_dbsnp = 0 if rs_ids is None else 1
rmsk_hits = annotations.get_rmsk_info(var)
in_cpg = annotations.get_cpg_island_info(var)
in_segdup = annotations.get_segdup_info(var)
is_conserved = annotations.get_conservation_info(var)
esp = annotations.get_esp_info(var)
thousandG = annotations.get_1000G_info(var)
recomb_rate = annotations.get_recomb_info(var)
gms = annotations.get_gms(var)
grc = annotations.get_grc(var)
in_cse = annotations.get_cse(var)
encode_tfbs = annotations.get_encode_tfbs(var)
encode_dnaseI = annotations.get_encode_dnase_clusters(var)
encode_cons_seg = annotations.get_encode_consensus_segs(var)
gerp_el = annotations.get_gerp_elements(var)
vista_enhancers = annotations.get_vista_enhancers(var)
cosmic_ids = annotations.get_cosmic_info(var)
fitcons = annotations.get_fitcons(var)
Exac = annotations.get_exac_info(var)
#load CADD scores by default
if self.args.skip_cadd is False:
(cadd_raw, cadd_scaled) = annotations.get_cadd_scores(var)
else:
(cadd_raw, cadd_scaled) = (None, None)
# load the GERP score for this variant by default.
gerp_bp = None
if self.args.skip_gerp_bp is False:
gerp_bp = annotations.get_gerp_bp(var)
# the variant is too big to annotate
else:
pfam_domain = None
cyto_band = None
rs_ids = None
clinvar_info = annotations.ClinVarInfo()
in_dbsnp = None
rmsk_hits = None
in_cpg = None
in_segdup = None
is_conserved = None
esp = annotations.ESPInfo(None, None, None, None, None)
thousandG = annotations.ThousandGInfo(None, None, None, None, None, None, None)
Exac = annotations.ExacInfo(None, None, None, None, None, None, None, None, None, None)
recomb_rate = None
gms = annotations.GmsTechs(None, None, None)
grc = None
in_cse = None
encode_tfbs = None
encode_dnaseI = annotations.ENCODEDnaseIClusters(None, None)
encode_cons_seg = annotations.ENCODESegInfo(None, None, None, None, None, None)
gerp_el = None
vista_enhancers = None
cosmic_ids = None
fitcons = None
cadd_raw = None
cadd_scaled = None
gerp_bp = None
# impact is a list of impacts for this variant
impacts = None
severe_impacts = None
# impact terms initialized to None for handling unannotated vcf's
# anno_id in variants is for the trans. with the most severe impact term
gene = transcript = exon = codon_change = aa_change = aa_length = \
biotype = consequence = consequence_so = effect_severity = None
is_coding = is_exonic = is_lof = None
polyphen_pred = polyphen_score = sift_pred = sift_score = anno_id = None
if self.args.anno_type is not None:
impacts = func_impact.interpret_impact(self.args, var, self._effect_fields)
il = [i for i in impacts if i.effect_severity]
# in case we don't have sever impact, we still try to get the impact
# to annote the main variants table.
if len(il) == 0 and len(impacts) > 0:
il = impacts[:1]
if len(il) > 0:
im = il[0]
transcript = im.transcript
exon, gene = im.exon, im.gene
effect_severity = im.effect_severity
codon_change = im.codon_change
biotype = im.biotype
is_coding = im.is_coding
aa_change, aa_length, consequence = im.aa_change, im.aa_length, im.consequence
sift_score = im.sift_score
polyphen_pred = im.polyphen_pred
polyphen_score = im.polyphen_score
sift_pred = im.sift_pred
sift_score = im.sift_score
anno_id = im.anno_id
is_exonic = im.is_exonic
is_coding = im.is_coding
is_lof = im.is_lof
severe_impacts = \
severe_impact.interpret_severe_impact(self.args, var, self._effect_fields)
if severe_impacts:
extra_fields.update(severe_impacts.extra_fields)
gene = severe_impacts.gene
transcript = severe_impacts.transcript
exon = severe_impacts.exon
codon_change = severe_impacts.codon_change
aa_change = severe_impacts.aa_change
aa_length = severe_impacts.aa_length
biotype = severe_impacts.biotype
consequence = severe_impacts.consequence
effect_severity = severe_impacts.effect_severity
polyphen_pred = severe_impacts.polyphen_pred
polyphen_score = severe_impacts.polyphen_score
sift_pred = severe_impacts.sift_pred
sift_score = severe_impacts.sift_score
anno_id = severe_impacts.anno_id
is_exonic = severe_impacts.is_exonic
is_coding = severe_impacts.is_coding
is_lof = severe_impacts.is_lof
consequence_so = severe_impacts.so
# construct the filter string
filter = None
if var.FILTER is not None and var.FILTER != ".":
if isinstance(var.FILTER, list):
filter = ";".join(var.FILTER)
else:
filter = var.FILTER
vcf_id = None
if var.ID is not None and var.ID != ".":
vcf_id = var.ID
# build up numpy arrays for the genotype information.
# these arrays will be pickled-to-binary, compressed,
# and loaded as SqlLite BLOB values (see compression.pack_blob)
gt_phred_ll_homref = gt_phred_ll_het = gt_phred_ll_homalt = None
if not self.args.no_genotypes and not self.args.no_load_genotypes:
gt_bases = np.array(var.gt_bases, np.str) # 'A/G', './.'
gt_types = np.array(var.gt_types, np.int8) # -1, 0, 1, 2
gt_phases = np.array(var.gt_phases, np.bool) # T F F
gt_depths = np.array(var.gt_depths, np.int32) # 10 37 0
gt_ref_depths = np.array(var.gt_ref_depths, np.int32) # 2 21 0 -1
gt_alt_depths = np.array(var.gt_alt_depths, np.int32) # 8 16 0 -1
gt_quals = np.array(var.gt_quals, np.float32) # 10.78 22 99 -1
gt_copy_numbers = np.array(var.gt_copy_numbers, np.float32) # 1.0 2.0 2.1 -1
gt_phred_likelihoods = get_phred_lik(var.gt_phred_likelihoods)
if gt_phred_likelihoods is not None:
gt_phred_ll_homref = gt_phred_likelihoods[:, 0]
gt_phred_ll_het = gt_phred_likelihoods[:, 1]
gt_phred_ll_homalt = gt_phred_likelihoods[:, 2]
# tally the genotypes
self._update_sample_gt_counts(gt_types)
else:
gt_bases = gt_types = gt_phases = gt_depths = gt_ref_depths = None
gt_alt_depths = gt_quals = gt_copy_numbers = None
if self.args.skip_info_string:
info = None
else:
info = var.INFO
# were functional impacts predicted by SnpEFF or VEP?
# if so, build up a row for each of the impacts / transcript
variant_impacts = []
for idx, impact in enumerate(impacts or [], start=1):
var_impact = [self.v_id, idx, impact.gene,
impact.transcript, impact.is_exonic,
impact.is_coding, impact.is_lof,
impact.exon, impact.codon_change,
impact.aa_change, impact.aa_length,
impact.biotype, impact.consequence,
impact.so, impact.effect_severity,
impact.polyphen_pred, impact.polyphen_score,
impact.sift_pred, impact.sift_score]
variant_impacts.append(var_impact)
# extract structural variants
sv = svs.StructuralVariant(var)
ci_left = sv.get_ci_left()
ci_right = sv.get_ci_right()
# construct the core variant record.
# 1 row per variant to VARIANTS table
chrom = var.CHROM if var.CHROM.startswith("chr") else "chr" + var.CHROM
variant = [chrom, var.start, var.end,
vcf_id, self.v_id, anno_id, var.REF, ','.join([x or "" for x in var.ALT]),
var.QUAL, filter, var.var_type,
var.var_subtype, pack_blob(gt_bases), pack_blob(gt_types),
pack_blob(gt_phases), pack_blob(gt_depths),
pack_blob(gt_ref_depths), pack_blob(gt_alt_depths),
pack_blob(gt_quals), pack_blob(gt_copy_numbers),
pack_blob(gt_phred_ll_homref),
pack_blob(gt_phred_ll_het),
pack_blob(gt_phred_ll_homalt),
call_rate, in_dbsnp,
rs_ids,
ci_left[0],
ci_left[1],
ci_right[0],
ci_right[1],
sv.get_length(),
sv.is_precise(),
sv.get_sv_tool(),
sv.get_evidence_type(),
sv.get_event_id(),
sv.get_mate_id(),
sv.get_strand(),
clinvar_info.clinvar_in_omim,
clinvar_info.clinvar_sig,
clinvar_info.clinvar_disease_name,
clinvar_info.clinvar_dbsource,
clinvar_info.clinvar_dbsource_id,
clinvar_info.clinvar_origin,
clinvar_info.clinvar_dsdb,
clinvar_info.clinvar_dsdbid,
clinvar_info.clinvar_disease_acc,
clinvar_info.clinvar_in_locus_spec_db,
clinvar_info.clinvar_on_diag_assay,
clinvar_info.clinvar_causal_allele,
pfam_domain, cyto_band, rmsk_hits, in_cpg,
in_segdup, is_conserved, gerp_bp, gerp_el,
hom_ref, het, hom_alt, unknown,
aaf, hwe_p_value, inbreeding_coeff, pi_hat,
recomb_rate, gene, transcript, is_exonic,
is_coding, is_lof, exon, codon_change, aa_change,
aa_length, biotype, consequence, consequence_so, effect_severity,
polyphen_pred, polyphen_score, sift_pred, sift_score,
infotag.get_ancestral_allele(var), infotag.get_rms_bq(var),
infotag.get_cigar(var),
infotag.get_depth(var), infotag.get_strand_bias(var),
infotag.get_rms_map_qual(var), infotag.get_homopol_run(var),
infotag.get_map_qual_zero(var),
infotag.get_num_of_alleles(var),
infotag.get_frac_dels(var),
infotag.get_haplotype_score(var),
infotag.get_quality_by_depth(var),
infotag.get_allele_count(var), infotag.get_allele_bal(var),
infotag.in_hm2(var), infotag.in_hm3(var),
infotag.is_somatic(var),
infotag.get_somatic_score(var),
esp.found, esp.aaf_EA,
esp.aaf_AA, esp.aaf_ALL,
esp.exome_chip, thousandG.found,
thousandG.aaf_AMR, thousandG.aaf_EAS, thousandG.aaf_SAS,
thousandG.aaf_AFR, thousandG.aaf_EUR,
thousandG.aaf_ALL, grc,
gms.illumina, gms.solid,
gms.iontorrent, in_cse,
encode_tfbs,
encode_dnaseI.cell_count,
encode_dnaseI.cell_list,
encode_cons_seg.gm12878,
encode_cons_seg.h1hesc,
encode_cons_seg.helas3,
encode_cons_seg.hepg2,
encode_cons_seg.huvec,
encode_cons_seg.k562,
vista_enhancers,
cosmic_ids,
pack_blob(info),
cadd_raw,
cadd_scaled,
fitcons,
Exac.found,
Exac.aaf_ALL,
Exac.adj_aaf_ALL,
Exac.aaf_AFR, Exac.aaf_AMR,
Exac.aaf_EAS, Exac.aaf_FIN,
Exac.aaf_NFE, Exac.aaf_OTH,
Exac.aaf_SAS]
return variant, variant_impacts, extra_fields
def _prepare_variation(self, var, anno_keys):
"""private method to collect metrics for a single variant (var) in a VCF file.
Extracts variant information, variant impacts and extra fields for annotation.
"""
extra_fields = {}
# these metric require that genotypes are present in the file
call_rate = None
hwe_p_value = None
pi_hat = None
inbreeding_coeff = None
hom_ref = het = hom_alt = unknown = None
# only compute certain metrics if genoypes are available
if not self.args.no_genotypes and not self.args.no_load_genotypes:
hom_ref = var.num_hom_ref
hom_alt = var.num_hom_alt
het = var.num_het
unknown = var.num_unknown
call_rate = var.call_rate
aaf = var.aaf
hwe_p_value, inbreeding_coeff = \
popgen.get_hwe_likelihood(hom_ref, het, hom_alt, aaf)
pi_hat = var.nucl_diversity
else:
aaf = infotag.extract_aaf(var)
if not isinstance(aaf, (float, int)):
if aaf is not None:
aaf = max(aaf)
############################################################
# collect annotations from gemini's custom annotation files
# but only if the size of the variant is <= 50kb
############################################################
if var.end - var.POS < 50000:
pfam_domain = annotations.get_pfamA_domains(var)
cyto_band = annotations.get_cyto_info(var)
rs_ids = annotations.get_dbsnp_info(var)
clinvar_info = annotations.get_clinvar_info(var)
in_dbsnp = 0 if rs_ids is None else 1
rmsk_hits = annotations.get_rmsk_info(var)
in_cpg = annotations.get_cpg_island_info(var)
in_segdup = annotations.get_segdup_info(var)
is_conserved = annotations.get_conservation_info(var)
esp = annotations.get_esp_info(var)
thousandG = annotations.get_1000G_info(var)
recomb_rate = annotations.get_recomb_info(var)
gms = annotations.get_gms(var)
grc = annotations.get_grc(var)
in_cse = annotations.get_cse(var)
encode_tfbs = annotations.get_encode_tfbs(var)
encode_dnaseI = annotations.get_encode_dnase_clusters(var)
encode_cons_seg = annotations.get_encode_consensus_segs(var)
gerp_el = annotations.get_gerp_elements(var)
vista_enhancers = annotations.get_vista_enhancers(var)
cosmic_ids = annotations.get_cosmic_info(var)
fitcons = annotations.get_fitcons(var)
Exac = annotations.get_exac_info(var)
#load CADD scores by default
if self.args.skip_cadd is False:
(cadd_raw, cadd_scaled) = annotations.get_cadd_scores(var)
else:
(cadd_raw, cadd_scaled) = (None, None)
# load the GERP score for this variant by default.
gerp_bp = None
if self.args.skip_gerp_bp is False:
gerp_bp = annotations.get_gerp_bp(var)
# the variant is too big to annotate
else:
pfam_domain = None
cyto_band = None
rs_ids = None
clinvar_info = annotations.ClinVarInfo()
in_dbsnp = None
rmsk_hits = None
in_cpg = None
in_segdup = None
is_conserved = None
esp = annotations.ESPInfo(False, -1, -1, -1, 0)
thousandG = annotations.EMPTY_1000G
Exac = annotations.EXAC_EMPTY
recomb_rate = None
gms = annotations.GmsTechs(None, None, None)
grc = None
in_cse = None
encode_tfbs = None
encode_dnaseI = annotations.ENCODEDnaseIClusters(None, None)
encode_cons_seg = annotations.ENCODESegInfo(None, None, None, None, None, None)
gerp_el = None
vista_enhancers = None
cosmic_ids = None
fitcons = None
cadd_raw = None
cadd_scaled = None
gerp_bp = None
top_impact = empty
if anno_keys == {}:
impacts = []
else:
impacts = []
if self.args.anno_type in ("all", "snpEff"):
try:
if "EFF" in anno_keys:
impacts += [geneimpacts.OldSnpEff(e, anno_keys["EFF"]) for e in var.INFO["EFF"].split(",")]
elif "ANN" in anno_keys:
impacts += [geneimpacts.SnpEff(e, anno_keys["ANN"]) for e in var.INFO["ANN"].split(",")]
except KeyError:
pass
if self.args.anno_type in ("all", "VEP"):
try:
impacts += [geneimpacts.VEP(e, anno_keys["CSQ"]) for e in var.INFO["CSQ"].split(",")]
except KeyError:
pass
for i, im in enumerate(impacts, start=1):
im.anno_id = i
if impacts != []:
top_impact = geneimpacts.Effect.top_severity(impacts)
if isinstance(top_impact, list):
top_impact = top_impact[0]
filter = None
if var.FILTER is not None and var.FILTER != ".":
if isinstance(var.FILTER, list):
filter = ";".join(var.FILTER)
else:
filter = var.FILTER
vcf_id = None
if var.ID is not None and var.ID != ".":
vcf_id = var.ID
chrom = var.CHROM if var.CHROM.startswith("chr") else "chr" + var.CHROM
clinvar_gene_phenotype = None
if top_impact.gene is not None:
clinvar_gene_phenotype = self.clinvar_chrom_gene_lookup.get((chrom[3:], top_impact.gene))
# build up numpy arrays for the genotype information.
# these arrays will be pickled-to-binary, compressed,
# and loaded as BLOB values (see compression.pack_blob)
gt_phred_ll_homref = gt_phred_ll_het = gt_phred_ll_homalt = None
if not (self.args.no_genotypes or self.args.no_load_genotypes):
gt_bases = var.gt_bases
gt_types = var.gt_types
gt_phases = var.gt_phases
gt_depths = var.gt_depths
gt_ref_depths = var.gt_ref_depths
gt_alt_depths = var.gt_alt_depths
gt_quals = var.gt_quals
#gt_copy_numbers = np.array(var.gt_copy_numbers, np.float32) # 1.0 2.0 2.1 -1
gt_copy_numbers = None
gt_phred_ll_homref = var.gt_phred_ll_homref
gt_phred_ll_het = var.gt_phred_ll_het
gt_phred_ll_homalt = var.gt_phred_ll_homalt
# tally the genotypes
self._update_sample_gt_counts(gt_types)
else:
gt_bases = gt_types = gt_phases = gt_depths = gt_ref_depths = None
gt_alt_depths = gt_quals = gt_copy_numbers = None
if self.args.skip_info_string:
info = None
else:
info = dict(var.INFO)
assert isinstance(thousandG.aaf_AMR, (int, float))
# were functional impacts predicted by SnpEFF or VEP?
# if so, build up a row for each of the impacts / transcript
variant_impacts = []
for idx, impact in enumerate(impacts or [], start=1):
var_impact = dict(variant_id=self.v_id, anno_id=idx, gene=impact.gene,
transcript=impact.transcript, is_exonic=impact.is_exonic,
is_coding=impact.is_coding, is_lof=impact.is_lof,
is_splicing=impact.is_splicing,
exon=impact.exon, codon_change=impact.codon_change,
aa_change=impact.aa_change, aa_length=impact.aa_length,
biotype=impact.biotype, impact=impact.top_consequence,
impact_so=impact.so, impact_severity=impact.effect_severity,
polyphed_pred=impact.polyphen_pred, polyphen_score=impact.polyphen_score,
sift_pred=impact.sift_pred,
sift_score=impact.sift_score)
variant_impacts.append(var_impact)
# extract structural variants
sv = svs.StructuralVariant(var)
ci_left = sv.get_ci_left()
ci_right = sv.get_ci_right()
if top_impact is not empty:
for dbkey, infokey in self._extra_effect_fields:
extra_fields[dbkey] = top_impact.effects[infokey]
if dbkey.endswith("_num"):
try:
extra_fields[dbkey] = float(extra_fields[dbkey])
except ValueError:
# sometimes the field is empty.
extra_fields[dbkey] = None
# construct the core variant record.
# 1 row per variant to VARIANTS table
variant = dict(chrom=chrom, start=var.start, end=var.end,
vcf_id=vcf_id, variant_id=self.v_id, anno_id=top_impact.anno_id,
ref=var.REF, alt=','.join([x or "" for x in var.ALT]),
qual=var.QUAL, filter=filter, type=var.var_type,
sub_type=var.var_subtype, gts=pack_blob(gt_bases),
gt_types=pack_blob(gt_types),
gt_phases=pack_blob(gt_phases), gt_depths=pack_blob(gt_depths),
gt_ref_depths=pack_blob(gt_ref_depths), gt_alt_depths=pack_blob(gt_alt_depths),
gt_quals=pack_blob(gt_quals), gt_copy_numbers=pack_blob(gt_copy_numbers),
gt_phred_ll_homref=pack_blob(gt_phred_ll_homref),
gt_phred_ll_het=pack_blob(gt_phred_ll_het),
gt_phred_ll_homalt=pack_blob(gt_phred_ll_homalt),
call_rate=call_rate, in_dbsnp=bool(in_dbsnp),
rs_ids=rs_ids,
sv_cipos_start_left=ci_left[0],
sv_cipos_end_left=ci_left[1],
sv_cipos_start_right=ci_right[0],
sv_cipos_end_right=ci_right[1],
sv_length=sv.get_length(),
sv_is_precise=sv.is_precise(),
sv_tool=sv.get_sv_tool(),
sv_evidence_type=sv.get_evidence_type(),
sv_event_id=sv.get_event_id(),
sv_mate_id=sv.get_mate_id(),
sv_strand=sv.get_strand(),
in_omim=bool(clinvar_info.clinvar_in_omim),
clinvar_sig=clinvar_info.clinvar_sig,
clinvar_disease_name=clinvar_info.clinvar_disease_name,
clinvar_dbsource=clinvar_info.clinvar_dbsource,
clinvar_dbsource_id=clinvar_info.clinvar_dbsource_id,
clinvar_origin=clinvar_info.clinvar_origin,
clinvar_dsdb=clinvar_info.clinvar_dsdb,
clinvar_dsdbid=clinvar_info.clinvar_dsdbid,
clinvar_disease_acc=clinvar_info.clinvar_disease_acc,
clinvar_in_locus_spec_db=bool(clinvar_info.clinvar_in_locus_spec_db),
clinvar_on_diag_assay=bool(clinvar_info.clinvar_on_diag_assay),
clinvar_causal_allele=clinvar_info.clinvar_causal_allele,
clinvar_gene_phenotype=clinvar_gene_phenotype,
geno2mp_hpo_ct=annotations.get_geno2mp_ct(var),
pfam_domain=pfam_domain, cyto_band=cyto_band, rmsk=rmsk_hits,
in_cpg_island=bool(in_cpg),
in_segdup=bool(in_segdup), is_conserved=bool(is_conserved),
gerp_bp_score=gerp_bp, gerp_element_pval=gerp_el,
num_hom_ref=hom_ref, num_het=het, num_hom_alt=hom_alt,
num_unknown=unknown,
aaf=aaf, hwe=hwe_p_value, inbreeding_coeff=inbreeding_coeff,
pi=pi_hat,
recomb_rate=recomb_rate, gene=top_impact.gene,
transcript=top_impact.transcript,
is_exonic=top_impact.is_exonic,
is_coding=top_impact.is_coding,
is_splicing=top_impact.is_splicing,
is_lof=top_impact.is_lof, exon=top_impact.exon,
codon_change=top_impact.codon_change, aa_change=top_impact.aa_change,
aa_length=top_impact.aa_length, biotype=top_impact.biotype,
impact=top_impact.top_consequence, impact_so=top_impact.so,
impact_severity=top_impact.effect_severity,
polyphen_pred=top_impact.polyphen_pred,
polyphen_score=top_impact.polyphen_score,
sift_pred=top_impact.sift_pred, sift_score=top_impact.sift_score,
anc_allele=infotag.get_ancestral_allele(var), rms_bq=infotag.get_rms_bq(var),
cigar=infotag.get_cigar(var),
depth=infotag.get_depth(var), strand_bias=infotag.get_strand_bias(var),
rms_map_qual=infotag.get_rms_map_qual(var), in_hom_run=infotag.get_homopol_run(var),
num_mapq_zero=infotag.get_map_qual_zero(var),
num_alleles=infotag.get_num_of_alleles(var),
num_reads_w_dels=infotag.get_frac_dels(var),
haplotype_score=infotag.get_haplotype_score(var),
qual_depth=infotag.get_quality_by_depth(var),
allele_count=infotag.get_allele_count(var), allele_bal=infotag.get_allele_bal(var),
# bools?
in_hm2=infotag.in_hm2(var), in_hm3=infotag.in_hm3(var),
is_somatic=infotag.is_somatic(var),
somatic_score=infotag.get_somatic_score(var),
in_esp=esp.found, aaf_esp_ea=esp.aaf_EA,
aaf_esp_aa=esp.aaf_AA, aaf_esp_all=esp.aaf_ALL,
exome_chip=bool(esp.exome_chip),
in_1kg=thousandG.found,
aaf_1kg_amr=thousandG.aaf_AMR,
aaf_1kg_eas=thousandG.aaf_EAS,
aaf_1kg_sas=thousandG.aaf_SAS,
aaf_1kg_afr=thousandG.aaf_AFR,
aaf_1kg_eur=thousandG.aaf_EUR,
aaf_1kg_all=thousandG.aaf_ALL,
grc=grc,
gms_illumina=gms.illumina,
gms_solid=gms.solid,
gms_iontorrent=gms.iontorrent, in_cse=in_cse,
encode_tfbs=encode_tfbs,
encode_dnaseI_cell_count=encode_dnaseI.cell_count,
encode_dnaseI_cell_list=encode_dnaseI.cell_list,
encode_consensus_gm12878=encode_cons_seg.gm12878,
encode_consensus_h1hesc=encode_cons_seg.h1hesc,
encode_consensus_helas3=encode_cons_seg.helas3,
encode_consensus_hepg2=encode_cons_seg.hepg2,
encode_consensus_huvec=encode_cons_seg.huvec,
encode_consensus_k562=encode_cons_seg.k562,
vista_enhancers=vista_enhancers,
cosmic_ids=cosmic_ids,
info=pack_blob(info),
cadd_raw=cadd_raw,
cadd_scaled=cadd_scaled,
fitcons=fitcons,
in_exac=Exac.found,
aaf_exac_all=Exac.aaf_ALL,
aaf_adj_exac_all=Exac.adj_aaf_ALL,
aaf_adj_exac_afr=Exac.aaf_AFR,
aaf_adj_exac_amr=Exac.aaf_AMR,
aaf_adj_exac_eas=Exac.aaf_EAS,
aaf_adj_exac_fin=Exac.aaf_FIN,
aaf_adj_exac_nfe=Exac.aaf_NFE,
aaf_adj_exac_oth=Exac.aaf_OTH,
aaf_adj_exac_sas=Exac.aaf_SAS,
exac_num_het=Exac.num_het,
exac_num_hom_alt=Exac.num_hom_alt,
exac_num_chroms=Exac.num_chroms)
variant['max_aaf_all'] = max(-1,
variant['aaf_esp_ea'],
variant['aaf_esp_aa'],
variant['aaf_1kg_amr'],
variant['aaf_1kg_eas'],
variant['aaf_1kg_sas'],
variant['aaf_1kg_afr'],
variant['aaf_1kg_eur'],
variant['aaf_adj_exac_afr'],
variant['aaf_adj_exac_amr'],
variant['aaf_adj_exac_eas'],
variant['aaf_adj_exac_nfe'],
variant['aaf_adj_exac_sas'])
variant.update(self._extra_empty)
return variant, variant_impacts, extra_fields
def _prepare_variation(self, var, anno_keys):
"""private method to collect metrics for a single variant (var) in a VCF file.
Extracts variant information, variant impacts and extra fields for annotation.
"""
extra_fields = {}
# these metric require that genotypes are present in the file
call_rate = None
hwe_p_value = None
pi_hat = None
inbreeding_coeff = None
hom_ref = het = hom_alt = unknown = None
# only compute certain metrics if genoypes are available
if not self.args.no_genotypes and not self.args.no_load_genotypes:
hom_ref = var.num_hom_ref
hom_alt = var.num_hom_alt
het = var.num_het
unknown = var.num_unknown
call_rate = var.call_rate
aaf = var.aaf
hwe_p_value, inbreeding_coeff = \
popgen.get_hwe_likelihood(hom_ref, het, hom_alt, aaf)
pi_hat = var.nucl_diversity
else:
aaf = infotag.extract_aaf(var)
if not isinstance(aaf, (float, int)):
if aaf is not None:
aaf = max(aaf)
############################################################
# collect annotations from gemini's custom annotation files
# but only if the size of the variant is <= 50kb
############################################################
if var.end - var.POS < 50000:
pfam_domain = annotations.get_pfamA_domains(var)
cyto_band = annotations.get_cyto_info(var)
rs_ids = annotations.get_dbsnp_info(var)
clinvar_info = annotations.get_clinvar_info(var)
in_dbsnp = 0 if rs_ids is None else 1
rmsk_hits = annotations.get_rmsk_info(var)
in_cpg = annotations.get_cpg_island_info(var)
in_segdup = annotations.get_segdup_info(var)
is_conserved = annotations.get_conservation_info(var)
esp = annotations.get_esp_info(var)
thousandG = annotations.get_1000G_info(var)
recomb_rate = annotations.get_recomb_info(var)
gms = annotations.get_gms(var)
grc = annotations.get_grc(var)
in_cse = annotations.get_cse(var)
encode_tfbs = annotations.get_encode_tfbs(var)
encode_dnaseI = annotations.get_encode_dnase_clusters(var)
encode_cons_seg = annotations.get_encode_consensus_segs(var)
gerp_el = annotations.get_gerp_elements(var)
vista_enhancers = annotations.get_vista_enhancers(var)
cosmic_ids = annotations.get_cosmic_info(var)
fitcons = annotations.get_fitcons(var)
Exac = annotations.get_exac_info(var)
#load CADD scores by default
if self.args.skip_cadd is False:
(cadd_raw, cadd_scaled) = annotations.get_cadd_scores(var)
else:
(cadd_raw, cadd_scaled) = (None, None)
# load the GERP score for this variant by default.
gerp_bp = None
if self.args.skip_gerp_bp is False:
gerp_bp = annotations.get_gerp_bp(var)
# the variant is too big to annotate
else:
pfam_domain = None
cyto_band = None
rs_ids = None
clinvar_info = annotations.ClinVarInfo()
in_dbsnp = None
rmsk_hits = None
in_cpg = None
in_segdup = None
is_conserved = None
esp = annotations.ESPInfo(None, None, None, None, None)
thousandG = annotations.ThousandGInfo(None, None, None, None, None, None, None)
Exac = annotations.ExacInfo(None, None, None, None, None, None,
None, None, None, None, None, None, None)
recomb_rate = None
gms = annotations.GmsTechs(None, None, None)
grc = None
in_cse = None
encode_tfbs = None
encode_dnaseI = annotations.ENCODEDnaseIClusters(None, None)
encode_cons_seg = annotations.ENCODESegInfo(None, None, None, None, None, None)
gerp_el = None
vista_enhancers = None
cosmic_ids = None
fitcons = None
cadd_raw = None
cadd_scaled = None
gerp_bp = None
top_impact = empty
if anno_keys == {}:
impacts = []
else:
impacts = []
if self.args.anno_type in ("all", "snpEff"):
try:
if "EFF" in anno_keys:
impacts += [geneimpacts.OldSnpEff(e, anno_keys["EFF"]) for e in var.INFO["EFF"].split(",")]
elif "ANN" in anno_keys:
impacts += [geneimpacts.SnpEff(e, anno_keys["ANN"]) for e in var.INFO["ANN"].split(",")]
except KeyError:
pass
elif self.args.anno_type in ("all", "VEP"):
try:
impacts += [geneimpacts.VEP(e, anno_keys["CSQ"]) for e in var.INFO["CSQ"].split(",")]
except KeyError:
pass
for i, im in enumerate(impacts, start=1):
im.anno_id = i
if impacts != []:
top_impact = geneimpacts.Effect.top_severity(impacts)
if isinstance(top_impact, list):
top_impact = top_impact[0]
filter = None
if var.FILTER is not None and var.FILTER != ".":
if isinstance(var.FILTER, list):
filter = ";".join(var.FILTER)
else:
filter = var.FILTER
vcf_id = None
if var.ID is not None and var.ID != ".":
vcf_id = var.ID
# build up numpy arrays for the genotype information.
# these arrays will be pickled-to-binary, compressed,
# and loaded as SqlLite BLOB values (see compression.pack_blob)
gt_phred_ll_homref = gt_phred_ll_het = gt_phred_ll_homalt = None
if not self.args.no_genotypes and not self.args.no_load_genotypes:
gt_bases = var.gt_bases
gt_types = var.gt_types
gt_phases = var.gt_phases
gt_depths = var.gt_depths
gt_ref_depths = var.gt_ref_depths
gt_alt_depths = var.gt_alt_depths
gt_quals = var.gt_quals
#gt_copy_numbers = np.array(var.gt_copy_numbers, np.float32) # 1.0 2.0 2.1 -1
gt_copy_numbers = None
gt_phred_ll_homref = var.gt_phred_ll_homref
gt_phred_ll_het = var.gt_phred_ll_het
gt_phred_ll_homalt = var.gt_phred_ll_homalt
# tally the genotypes
self._update_sample_gt_counts(gt_types)
else:
gt_bases = gt_types = gt_phases = gt_depths = gt_ref_depths = None
gt_alt_depths = gt_quals = gt_copy_numbers = None
if self.args.skip_info_string:
info = None
else:
info = dict(var.INFO)
# were functional impacts predicted by SnpEFF or VEP?
# if so, build up a row for each of the impacts / transcript
variant_impacts = []
for idx, impact in enumerate(impacts or [], start=1):
var_impact = [self.v_id, idx, impact.gene,
impact.transcript, impact.is_exonic,
impact.is_coding,
impact.is_splicing,
impact.is_lof,
impact.exon, impact.codon_change,
impact.aa_change, impact.aa_length,
impact.biotype, impact.top_consequence,
impact.so, impact.effect_severity,
impact.polyphen_pred, impact.polyphen_score,
impact.sift_pred, impact.sift_score]
variant_impacts.append(var_impact)
# extract structural variants
sv = svs.StructuralVariant(var)
ci_left = sv.get_ci_left()
ci_right = sv.get_ci_right()
if top_impact is not empty:
for dbkey, infokey in self._extra_effect_fields:
extra_fields[dbkey] = top_impact.effects[infokey]
# construct the core variant record.
# 1 row per variant to VARIANTS table
chrom = var.CHROM if var.CHROM.startswith("chr") else "chr" + var.CHROM
variant = [chrom, var.start, var.end,
vcf_id, self.v_id, top_impact.anno_id, var.REF, ','.join([x or "" for x in var.ALT]),
var.QUAL, filter, var.var_type,
var.var_subtype, pack_blob(gt_bases), pack_blob(gt_types),
pack_blob(gt_phases), pack_blob(gt_depths),
pack_blob(gt_ref_depths), pack_blob(gt_alt_depths),
pack_blob(gt_quals), pack_blob(gt_copy_numbers),
pack_blob(gt_phred_ll_homref),
pack_blob(gt_phred_ll_het),
pack_blob(gt_phred_ll_homalt),
call_rate, in_dbsnp,
rs_ids,
ci_left[0],
ci_left[1],
ci_right[0],
ci_right[1],
sv.get_length(),
sv.is_precise(),
sv.get_sv_tool(),
sv.get_evidence_type(),
sv.get_event_id(),
sv.get_mate_id(),
sv.get_strand(),
clinvar_info.clinvar_in_omim,
clinvar_info.clinvar_sig,
clinvar_info.clinvar_disease_name,
clinvar_info.clinvar_dbsource,
clinvar_info.clinvar_dbsource_id,
clinvar_info.clinvar_origin,
clinvar_info.clinvar_dsdb,
clinvar_info.clinvar_dsdbid,
clinvar_info.clinvar_disease_acc,
clinvar_info.clinvar_in_locus_spec_db,
clinvar_info.clinvar_on_diag_assay,
clinvar_info.clinvar_causal_allele,
pfam_domain, cyto_band, rmsk_hits, in_cpg,
in_segdup, is_conserved, gerp_bp, gerp_el,
hom_ref, het, hom_alt, unknown,
aaf, hwe_p_value, inbreeding_coeff, pi_hat,
recomb_rate,
top_impact.gene,
top_impact.transcript,
top_impact.is_exonic,
top_impact.is_coding,
top_impact.is_splicing,
top_impact.is_lof,
top_impact.exon,
top_impact.codon_change,
top_impact.aa_change,
top_impact.aa_length,
top_impact.biotype,
top_impact.top_consequence,
top_impact.so,
top_impact.effect_severity,
top_impact.polyphen_pred,
top_impact.polyphen_score,
top_impact.sift_pred,
top_impact.sift_score,
infotag.get_ancestral_allele(var), infotag.get_rms_bq(var),
infotag.get_cigar(var),
infotag.get_depth(var), infotag.get_strand_bias(var),
infotag.get_rms_map_qual(var), infotag.get_homopol_run(var),
infotag.get_map_qual_zero(var),
infotag.get_num_of_alleles(var),
infotag.get_frac_dels(var),
infotag.get_haplotype_score(var),
infotag.get_quality_by_depth(var),
infotag.get_allele_count(var), infotag.get_allele_bal(var),
infotag.in_hm2(var), infotag.in_hm3(var),
infotag.is_somatic(var),
infotag.get_somatic_score(var),
esp.found, esp.aaf_EA,
esp.aaf_AA, esp.aaf_ALL,
esp.exome_chip, thousandG.found,
thousandG.aaf_AMR, thousandG.aaf_EAS, thousandG.aaf_SAS,
thousandG.aaf_AFR, thousandG.aaf_EUR,
thousandG.aaf_ALL, grc,
gms.illumina, gms.solid,
gms.iontorrent, in_cse,
encode_tfbs,
encode_dnaseI.cell_count,
encode_dnaseI.cell_list,
encode_cons_seg.gm12878,
encode_cons_seg.h1hesc,
encode_cons_seg.helas3,
encode_cons_seg.hepg2,
encode_cons_seg.huvec,
encode_cons_seg.k562,
vista_enhancers,
cosmic_ids,
pack_blob(info),
cadd_raw,
cadd_scaled,
fitcons,
Exac.found,
Exac.aaf_ALL,
Exac.adj_aaf_ALL,
Exac.aaf_AFR, Exac.aaf_AMR,
Exac.aaf_EAS, Exac.aaf_FIN,
Exac.aaf_NFE, Exac.aaf_OTH,
Exac.aaf_SAS,
Exac.num_het,
Exac.num_hom_alt,
Exac.num_chroms]
return variant, variant_impacts, extra_fields
def _prepare_variation(self, var):
"""private method to collect metrics for a single variant (var) in a VCF file.
Extracts variant information, variant impacts and extra fields for annotation.
"""
extra_fields = {}
# these metric require that genotypes are present in the file
call_rate = None
hwe_p_value = None
pi_hat = None
inbreeding_coeff = None
hom_ref = het = hom_alt = unknown = None
# only compute certain metrics if genoypes are available
if not self.args.no_genotypes and not self.args.no_load_genotypes:
hom_ref = var.num_hom_ref
hom_alt = var.num_hom_alt
het = var.num_het
unknown = var.num_unknown
call_rate = var.call_rate
aaf = var.aaf
hwe_p_value, inbreeding_coeff = \
popgen.get_hwe_likelihood(hom_ref, het, hom_alt, aaf)
pi_hat = var.nucl_diversity
else:
aaf = infotag.extract_aaf(var)
############################################################
# collect annotations from gemini's custom annotation files
# but only if the size of the variant is <= 50kb
############################################################
if var.end - var.POS < 50000:
pfam_domain = annotations.get_pfamA_domains(var)
cyto_band = annotations.get_cyto_info(var)
rs_ids = annotations.get_dbsnp_info(var)
clinvar_info = annotations.get_clinvar_info(var)
in_dbsnp = 0 if rs_ids is None else 1
rmsk_hits = annotations.get_rmsk_info(var)
in_cpg = annotations.get_cpg_island_info(var)
in_segdup = annotations.get_segdup_info(var)
is_conserved = annotations.get_conservation_info(var)
esp = annotations.get_esp_info(var)
thousandG = annotations.get_1000G_info(var)
recomb_rate = annotations.get_recomb_info(var)
gms = annotations.get_gms(var)
grc = annotations.get_grc(var)
in_cse = annotations.get_cse(var)
encode_tfbs = annotations.get_encode_tfbs(var)
encode_dnaseI = annotations.get_encode_dnase_clusters(var)
encode_cons_seg = annotations.get_encode_consensus_segs(var)
gerp_el = annotations.get_gerp_elements(var)
vista_enhancers = annotations.get_vista_enhancers(var)
cosmic_ids = annotations.get_cosmic_info(var)
fitcons = annotations.get_fitcons(var)
Exac = annotations.get_exac_info(var)
#load CADD scores by default
if self.args.skip_cadd is False:
(cadd_raw, cadd_scaled) = annotations.get_cadd_scores(var)
else:
(cadd_raw, cadd_scaled) = (None, None)
# load the GERP score for this variant by default.
gerp_bp = None
if self.args.skip_gerp_bp is False:
gerp_bp = annotations.get_gerp_bp(var)
# the variant is too big to annotate
else:
pfam_domain = None
cyto_band = None
rs_ids = None
clinvar_info = annotations.ClinVarInfo()
in_dbsnp = None
rmsk_hits = None
in_cpg = None
in_segdup = None
is_conserved = None
esp = annotations.ESPInfo(None, None, None, None, None)
thousandG = annotations.ThousandGInfo(None, None, None, None, None,
None, None)
Exac = annotations.ExacInfo(None, None, None, None, None, None,
None, None, None, None)
recomb_rate = None
gms = annotations.GmsTechs(None, None, None)
grc = None
in_cse = None
encode_tfbs = None
encode_dnaseI = annotations.ENCODEDnaseIClusters(None, None)
encode_cons_seg = annotations.ENCODESegInfo(
None, None, None, None, None, None)
gerp_el = None
vista_enhancers = None
cosmic_ids = None
fitcons = None
cadd_raw = None
cadd_scaled = None
gerp_bp = None
# impact is a list of impacts for this variant
impacts = None
severe_impacts = None
# impact terms initialized to None for handling unannotated vcf's
# anno_id in variants is for the trans. with the most severe impact term
gene = transcript = exon = codon_change = aa_change = aa_length = \
biotype = consequence = consequence_so = effect_severity = None
is_coding = is_exonic = is_lof = None
polyphen_pred = polyphen_score = sift_pred = sift_score = anno_id = None
if self.args.anno_type is not None:
impacts = func_impact.interpret_impact(self.args, var,
self._effect_fields)
severe_impacts = \
severe_impact.interpret_severe_impact(self.args, var, self._effect_fields)
if severe_impacts:
extra_fields.update(severe_impacts.extra_fields)
gene = severe_impacts.gene
transcript = severe_impacts.transcript
exon = severe_impacts.exon
codon_change = severe_impacts.codon_change
aa_change = severe_impacts.aa_change
aa_length = severe_impacts.aa_length
biotype = severe_impacts.biotype
consequence = severe_impacts.consequence
effect_severity = severe_impacts.effect_severity
polyphen_pred = severe_impacts.polyphen_pred
polyphen_score = severe_impacts.polyphen_score
sift_pred = severe_impacts.sift_pred
sift_score = severe_impacts.sift_score
anno_id = severe_impacts.anno_id
is_exonic = severe_impacts.is_exonic
is_coding = severe_impacts.is_coding
is_lof = severe_impacts.is_lof
consequence_so = severe_impacts.so
# construct the filter string
filter = None
if var.FILTER is not None and var.FILTER != ".":
if isinstance(var.FILTER, list):
filter = ";".join(var.FILTER)
else:
filter = var.FILTER
vcf_id = None
if var.ID is not None and var.ID != ".":
vcf_id = var.ID
# build up numpy arrays for the genotype information.
# these arrays will be pickled-to-binary, compressed,
# and loaded as SqlLite BLOB values (see compression.pack_blob)
if not self.args.no_genotypes and not self.args.no_load_genotypes:
gt_bases = np.array(var.gt_bases, np.str) # 'A/G', './.'
gt_types = np.array(var.gt_types, np.int8) # -1, 0, 1, 2
gt_phases = np.array(var.gt_phases, np.bool) # T F F
gt_depths = np.array(var.gt_depths, np.int32) # 10 37 0
gt_ref_depths = np.array(var.gt_ref_depths, np.int32) # 2 21 0 -1
gt_alt_depths = np.array(var.gt_alt_depths, np.int32) # 8 16 0 -1
gt_quals = np.array(var.gt_quals, np.float32) # 10.78 22 99 -1
gt_copy_numbers = np.array(var.gt_copy_numbers,
np.float32) # 1.0 2.0 2.1 -1
# tally the genotypes
self._update_sample_gt_counts(gt_types)
else:
gt_bases = None
gt_types = None
gt_phases = None
gt_depths = None
gt_ref_depths = None
gt_alt_depths = None
gt_quals = None
gt_copy_numbers = None
if self.args.skip_info_string is False:
info = var.INFO
else:
info = None
# were functional impacts predicted by SnpEFF or VEP?
# if so, build up a row for each of the impacts / transcript
variant_impacts = []
if impacts is not None:
for idx, impact in enumerate(impacts):
var_impact = [
self.v_id, (idx + 1), impact.gene, impact.transcript,
impact.is_exonic, impact.is_coding, impact.is_lof,
impact.exon, impact.codon_change, impact.aa_change,
impact.aa_length, impact.biotype, impact.consequence,
impact.so, impact.effect_severity, impact.polyphen_pred,
impact.polyphen_score, impact.sift_pred, impact.sift_score
]
variant_impacts.append(var_impact)
# extract structural variants
sv = svs.StructuralVariant(var)
ci_left = sv.get_ci_left()
ci_right = sv.get_ci_right()
# construct the core variant record.
# 1 row per variant to VARIANTS table
if extra_fields:
extra_fields.update({
"chrom": var.CHROM,
"start": var.start,
"end": var.end
})
chrom = var.CHROM if var.CHROM.startswith("chr") else "chr" + var.CHROM
variant = [
chrom, var.start, var.end, vcf_id, self.v_id, anno_id, var.REF,
','.join([x or "" for x in var.ALT
]), var.QUAL, filter, var.var_type, var.var_subtype,
pack_blob(gt_bases),
pack_blob(gt_types),
pack_blob(gt_phases),
pack_blob(gt_depths),
pack_blob(gt_ref_depths),
pack_blob(gt_alt_depths),
pack_blob(gt_quals),
pack_blob(gt_copy_numbers), call_rate, in_dbsnp, rs_ids,
ci_left[0], ci_left[1], ci_right[0], ci_right[1],
sv.get_length(),
sv.is_precise(),
sv.get_sv_tool(),
sv.get_evidence_type(),
sv.get_event_id(),
sv.get_mate_id(),
sv.get_strand(), clinvar_info.clinvar_in_omim,
clinvar_info.clinvar_sig, clinvar_info.clinvar_disease_name,
clinvar_info.clinvar_dbsource, clinvar_info.clinvar_dbsource_id,
clinvar_info.clinvar_origin, clinvar_info.clinvar_dsdb,
clinvar_info.clinvar_dsdbid, clinvar_info.clinvar_disease_acc,
clinvar_info.clinvar_in_locus_spec_db,
clinvar_info.clinvar_on_diag_assay,
clinvar_info.clinvar_causal_allele, pfam_domain, cyto_band,
rmsk_hits, in_cpg, in_segdup, is_conserved, gerp_bp, gerp_el,
hom_ref, het, hom_alt, unknown, aaf, hwe_p_value, inbreeding_coeff,
pi_hat, recomb_rate, gene, transcript, is_exonic, is_coding,
is_lof, exon, codon_change, aa_change, aa_length, biotype,
consequence, consequence_so, effect_severity, polyphen_pred,
polyphen_score, sift_pred, sift_score,
infotag.get_ancestral_allele(var),
infotag.get_rms_bq(var),
infotag.get_cigar(var),
infotag.get_depth(var),
infotag.get_strand_bias(var),
infotag.get_rms_map_qual(var),
infotag.get_homopol_run(var),
infotag.get_map_qual_zero(var),
infotag.get_num_of_alleles(var),
infotag.get_frac_dels(var),
infotag.get_haplotype_score(var),
infotag.get_quality_by_depth(var),
infotag.get_allele_count(var),
infotag.get_allele_bal(var),
infotag.in_hm2(var),
infotag.in_hm3(var),
infotag.is_somatic(var),
infotag.get_somatic_score(var), esp.found, esp.aaf_EA, esp.aaf_AA,
esp.aaf_ALL, esp.exome_chip, thousandG.found, thousandG.aaf_AMR,
thousandG.aaf_EAS, thousandG.aaf_SAS, thousandG.aaf_AFR,
thousandG.aaf_EUR, thousandG.aaf_ALL, grc, gms.illumina, gms.solid,
gms.iontorrent, in_cse, encode_tfbs, encode_dnaseI.cell_count,
encode_dnaseI.cell_list, encode_cons_seg.gm12878,
encode_cons_seg.h1hesc, encode_cons_seg.helas3,
encode_cons_seg.hepg2, encode_cons_seg.huvec, encode_cons_seg.k562,
vista_enhancers, cosmic_ids,
pack_blob(info), cadd_raw, cadd_scaled, fitcons, Exac.found,
Exac.aaf_ALL, Exac.adj_aaf_ALL, Exac.aaf_AFR, Exac.aaf_AMR,
Exac.aaf_EAS, Exac.aaf_FIN, Exac.aaf_NFE, Exac.aaf_OTH,
Exac.aaf_SAS
]
return variant, variant_impacts, extra_fields
def _prepare_variation(self, var, anno_keys):
"""private method to collect metrics for a single variant (var) in a VCF file.
Extracts variant information, variant impacts and extra fields for annotation.
"""
extra_fields = {}
# these metric require that genotypes are present in the file
call_rate = None
hwe_p_value = None
pi_hat = None
inbreeding_coeff = None
hom_ref = het = hom_alt = unknown = None
# only compute certain metrics if genoypes are available
if not self.args.no_genotypes and not self.args.no_load_genotypes:
hom_ref = var.num_hom_ref
hom_alt = var.num_hom_alt
het = var.num_het
unknown = var.num_unknown
call_rate = var.call_rate
aaf = var.aaf
hwe_p_value, inbreeding_coeff = \
popgen.get_hwe_likelihood(hom_ref, het, hom_alt, aaf)
pi_hat = var.nucl_diversity
else:
aaf = infotag.extract_aaf(var)
if not isinstance(aaf, (float, int)):
if aaf is not None:
aaf = max(aaf)
############################################################
# collect annotations from gemini's custom annotation files
# but only if the size of the variant is <= 50kb
############################################################
if var.end - var.POS < 50000:
pfam_domain = annotations.get_pfamA_domains(var)
cyto_band = annotations.get_cyto_info(var)
rs_ids = annotations.get_dbsnp_info(var)
clinvar_info = annotations.get_clinvar_info(var)
in_dbsnp = 0 if rs_ids is None else 1
rmsk_hits = annotations.get_rmsk_info(var)
in_cpg = annotations.get_cpg_island_info(var)
in_segdup = annotations.get_segdup_info(var)
is_conserved = annotations.get_conservation_info(var)
esp = annotations.get_esp_info(var)
thousandG = annotations.get_1000G_info(var)
recomb_rate = annotations.get_recomb_info(var)
gms = annotations.get_gms(var)
grc = annotations.get_grc(var)
in_cse = annotations.get_cse(var)
encode_tfbs = annotations.get_encode_tfbs(var)
encode_dnaseI = annotations.get_encode_dnase_clusters(var)
encode_cons_seg = annotations.get_encode_consensus_segs(var)
gerp_el = annotations.get_gerp_elements(var)
vista_enhancers = annotations.get_vista_enhancers(var)
cosmic_ids = annotations.get_cosmic_info(var)
fitcons = annotations.get_fitcons(var)
Exac = annotations.get_exac_info(var)
#load CADD scores by default
if self.args.skip_cadd is False:
(cadd_raw, cadd_scaled) = annotations.get_cadd_scores(var)
else:
(cadd_raw, cadd_scaled) = (None, None)
# load the GERP score for this variant by default.
gerp_bp = None
if self.args.skip_gerp_bp is False:
gerp_bp = annotations.get_gerp_bp(var)
# the variant is too big to annotate
else:
pfam_domain = None
cyto_band = None
rs_ids = None
clinvar_info = annotations.ClinVarInfo()
in_dbsnp = None
rmsk_hits = None
in_cpg = None
in_segdup = None
is_conserved = None
esp = annotations.ESPInfo(None, None, None, None, None)
thousandG = annotations.ThousandGInfo(None, None, None, None, None,
None, None)
Exac = annotations.ExacInfo(None, None, None, None, None, None,
None, None, None, None, None, None,
None)
recomb_rate = None
gms = annotations.GmsTechs(None, None, None)
grc = None
in_cse = None
encode_tfbs = None
encode_dnaseI = annotations.ENCODEDnaseIClusters(None, None)
encode_cons_seg = annotations.ENCODESegInfo(
None, None, None, None, None, None)
gerp_el = None
vista_enhancers = None
cosmic_ids = None
fitcons = None
cadd_raw = None
cadd_scaled = None
gerp_bp = None
top_impact = empty
if anno_keys == {}:
impacts = []
else:
impacts = []
if self.args.anno_type in ("all", "snpEff"):
try:
if "EFF" in anno_keys:
impacts += [
geneimpacts.OldSnpEff(e, anno_keys["EFF"])
for e in var.INFO["EFF"].split(",")
]
elif "ANN" in anno_keys:
impacts += [
geneimpacts.SnpEff(e, anno_keys["ANN"])
for e in var.INFO["ANN"].split(",")
]
except KeyError:
pass
if self.args.anno_type in ("all", "VEP"):
try:
impacts += [
geneimpacts.VEP(e, anno_keys["CSQ"])
for e in var.INFO["CSQ"].split(",")
]
except KeyError:
pass
for i, im in enumerate(impacts, start=1):
im.anno_id = i
if impacts != []:
top_impact = geneimpacts.Effect.top_severity(impacts)
if isinstance(top_impact, list):
top_impact = top_impact[0]
filter = None
if var.FILTER is not None and var.FILTER != ".":
if isinstance(var.FILTER, list):
filter = ";".join(var.FILTER)
else:
filter = var.FILTER
vcf_id = None
if var.ID is not None and var.ID != ".":
vcf_id = var.ID
chrom = var.CHROM if var.CHROM.startswith("chr") else "chr" + var.CHROM
clinvar_gene_phenotype = None
if top_impact.gene is not None:
clinvar_gene_phenotype = self.clinvar_chrom_gene_lookup.get(
(chrom[3:], top_impact.gene))
# build up numpy arrays for the genotype information.
# these arrays will be pickled-to-binary, compressed,
# and loaded as BLOB values (see compression.pack_blob)
gt_phred_ll_homref = gt_phred_ll_het = gt_phred_ll_homalt = None
if not self.args.no_genotypes and not self.args.no_load_genotypes:
gt_bases = var.gt_bases
gt_types = var.gt_types
gt_phases = var.gt_phases
gt_depths = var.gt_depths
gt_ref_depths = var.gt_ref_depths
gt_alt_depths = var.gt_alt_depths
gt_quals = var.gt_quals
#gt_copy_numbers = np.array(var.gt_copy_numbers, np.float32) # 1.0 2.0 2.1 -1
gt_copy_numbers = None
gt_phred_ll_homref = var.gt_phred_ll_homref
gt_phred_ll_het = var.gt_phred_ll_het
gt_phred_ll_homalt = var.gt_phred_ll_homalt
# tally the genotypes
self._update_sample_gt_counts(gt_types)
else:
gt_bases = gt_types = gt_phases = gt_depths = gt_ref_depths = None
gt_alt_depths = gt_quals = gt_copy_numbers = None
if self.args.skip_info_string:
info = None
else:
info = dict(var.INFO)
# were functional impacts predicted by SnpEFF or VEP?
# if so, build up a row for each of the impacts / transcript
variant_impacts = []
for idx, impact in enumerate(impacts or [], start=1):
var_impact = [
self.v_id, idx, impact.gene, impact.transcript,
impact.is_exonic, impact.is_coding, impact.is_splicing,
impact.is_lof, impact.exon, impact.codon_change,
impact.aa_change, impact.aa_length, impact.biotype,
impact.top_consequence, impact.so, impact.effect_severity,
impact.polyphen_pred, impact.polyphen_score, impact.sift_pred,
impact.sift_score
]
variant_impacts.append(var_impact)
# extract structural variants
sv = svs.StructuralVariant(var)
ci_left = sv.get_ci_left()
ci_right = sv.get_ci_right()
if top_impact is not empty:
for dbkey, infokey in self._extra_effect_fields:
extra_fields[dbkey] = top_impact.effects[infokey]
# construct the core variant record.
# 1 row per variant to VARIANTS table
variant = [
chrom, var.start, var.end, vcf_id, self.v_id, top_impact.anno_id,
var.REF, ','.join([x or "" for x in var.ALT]), var.QUAL, filter,
var.var_type, var.var_subtype,
pack_blob(gt_bases),
pack_blob(gt_types),
pack_blob(gt_phases),
pack_blob(gt_depths),
pack_blob(gt_ref_depths),
pack_blob(gt_alt_depths),
pack_blob(gt_quals),
pack_blob(gt_copy_numbers),
pack_blob(gt_phred_ll_homref),
pack_blob(gt_phred_ll_het),
pack_blob(gt_phred_ll_homalt), call_rate, in_dbsnp, rs_ids,
ci_left[0], ci_left[1], ci_right[0], ci_right[1],
sv.get_length(),
sv.is_precise(),
sv.get_sv_tool(),
sv.get_evidence_type(),
sv.get_event_id(),
sv.get_mate_id(),
sv.get_strand(), clinvar_info.clinvar_in_omim,
clinvar_info.clinvar_sig, clinvar_info.clinvar_disease_name,
clinvar_info.clinvar_dbsource, clinvar_info.clinvar_dbsource_id,
clinvar_info.clinvar_origin, clinvar_info.clinvar_dsdb,
clinvar_info.clinvar_dsdbid, clinvar_info.clinvar_disease_acc,
clinvar_info.clinvar_in_locus_spec_db,
clinvar_info.clinvar_on_diag_assay,
clinvar_info.clinvar_causal_allele, clinvar_gene_phenotype,
pfam_domain, cyto_band, rmsk_hits, in_cpg, in_segdup, is_conserved,
gerp_bp, gerp_el, hom_ref, het, hom_alt, unknown, aaf, hwe_p_value,
inbreeding_coeff, pi_hat, recomb_rate, top_impact.gene,
top_impact.transcript, top_impact.is_exonic, top_impact.is_coding,
top_impact.is_splicing, top_impact.is_lof, top_impact.exon,
top_impact.codon_change, top_impact.aa_change,
top_impact.aa_length, top_impact.biotype,
top_impact.top_consequence, top_impact.so,
top_impact.effect_severity, top_impact.polyphen_pred,
top_impact.polyphen_score, top_impact.sift_pred,
top_impact.sift_score,
infotag.get_ancestral_allele(var),
infotag.get_rms_bq(var),
infotag.get_cigar(var),
infotag.get_depth(var),
infotag.get_strand_bias(var),
infotag.get_rms_map_qual(var),
infotag.get_homopol_run(var),
infotag.get_map_qual_zero(var),
infotag.get_num_of_alleles(var),
infotag.get_frac_dels(var),
infotag.get_haplotype_score(var),
infotag.get_quality_by_depth(var),
infotag.get_allele_count(var),
infotag.get_allele_bal(var),
infotag.in_hm2(var),
infotag.in_hm3(var),
infotag.is_somatic(var),
infotag.get_somatic_score(var), esp.found, esp.aaf_EA, esp.aaf_AA,
esp.aaf_ALL, esp.exome_chip, thousandG.found, thousandG.aaf_AMR,
thousandG.aaf_EAS, thousandG.aaf_SAS, thousandG.aaf_AFR,
thousandG.aaf_EUR, thousandG.aaf_ALL, grc, gms.illumina, gms.solid,
gms.iontorrent, in_cse, encode_tfbs, encode_dnaseI.cell_count,
encode_dnaseI.cell_list, encode_cons_seg.gm12878,
encode_cons_seg.h1hesc, encode_cons_seg.helas3,
encode_cons_seg.hepg2, encode_cons_seg.huvec, encode_cons_seg.k562,
vista_enhancers, cosmic_ids,
pack_blob(info), cadd_raw, cadd_scaled, fitcons, Exac.found,
Exac.aaf_ALL, Exac.adj_aaf_ALL, Exac.aaf_AFR, Exac.aaf_AMR,
Exac.aaf_EAS, Exac.aaf_FIN, Exac.aaf_NFE, Exac.aaf_OTH,
Exac.aaf_SAS, Exac.num_het, Exac.num_hom_alt, Exac.num_chroms
]
return variant, variant_impacts, extra_fields