input_vds_path = str(args.input_vds).rstrip("/")
if not input_vds_path.endswith(".vds"):
p.error("Input must be a .vds")
input_vds_path_prefix = input_vds_path.replace(".vds", "")
logger.info("\n==> create HailContext")
hc = hail.HailContext(log="/hail.log")
logger.info("\n==> import vds: " + input_vds_path)
vds = hc.read(input_vds_path)
parallel_computed_annotation_exprs = [
"va.variantId = %s" % get_expr_for_variant_id(),
"va.contig = %s" % get_expr_for_contig(),
"va.start = %s" % get_expr_for_start_pos(),
"va.pos = %s" % get_expr_for_start_pos(),
"va.end = %s" % get_expr_for_end_pos(),
"va.ref = %s" % get_expr_for_ref_allele(),
"va.alt = %s" % get_expr_for_alt_allele(),
"va.xpos = %s" % get_expr_for_xpos(pos_field="start"),
"va.xstart = %s" % get_expr_for_xpos(pos_field="start"),
]
serial_computed_annotation_exprs = [
"va.xstop = %s" % get_expr_for_xpos(field_prefix="va.", pos_field="end"),
]
vds = vds.annotate_variants_expr(parallel_computed_annotation_exprs)
vds = vds.annotate_variants_expr(serial_computed_annotation_exprs)
NCC: Int,
NEGATIVE_TRAIN_SITE: Boolean,
POSITIVE_TRAIN_SITE: Boolean,
QD: Double,
ReadPosRankSum: Double,
SOR: Double,
VQSLOD: Double,
culprit: String,
AC_Hom: Array[Int],
AC_Het: Array[Int],
AC_Hemi: Array[Int],
"""
}
vds_computed_annotations_exprs = [
"va.chrom = %s" % get_expr_for_contig(),
"va.pos = %s" % get_expr_for_start_pos(),
"va.ref = %s" % get_expr_for_ref_allele(),
"va.alt = %s" % get_expr_for_alt_allele(),
"va.xpos = %s" % get_expr_for_xpos(),
"va.variantId = %s" % get_expr_for_variant_id(),
"va.originalAltAlleles = %s" % get_expr_for_orig_alt_alleles_set(),
"va.geneIds = %s" % get_expr_for_vep_gene_ids_set(),
"va.transcriptIds = %s" % get_expr_for_vep_transcript_ids_set(),
"va.transcriptConsequenceTerms = %s" % get_expr_for_vep_consequence_terms_set(),
"va.sortedTranscriptConsequences = %s" % get_expr_for_vep_sorted_transcript_consequences_array(),
"va.mainTranscript = %s" % get_expr_for_worst_transcript_consequence_annotations_struct("va.sortedTranscriptConsequences"),
"va.sortedTranscriptConsequences = json(va.sortedTranscriptConsequences)"
]
pprint(kt_pop.schema)
pprint(kt_annotations.schema)
pprint(kt_rare_variants.schema)
ES_HOST_IP = '10.4.0.13'
ES_HOST_PORT = 9200
print("======== Export to elasticsearch ======")
es = ElasticsearchClient(
host=ES_HOST_IP,
port=ES_HOST_PORT,
)
annotation_expressions = [
'variant_id = %s' % get_expr_for_variant_id(),
'chrom = %s' % get_expr_for_contig(),
'pos = %s' % get_expr_for_start_pos(),
"xpos = %s" % get_expr_for_xpos(field_prefix="", pos_field="pos"),
]
for expression in annotation_expressions:
kt_rare_variants = kt_rare_variants.annotate(expression)
kt_rare_variants = kt_rare_variants.drop(['v'])
kt_annotations = kt_annotations.annotate('variantId = %s' % get_expr_for_variant_id()).drop(['v'])
kt_rare_variants = kt_rare_variants.key_by('variantId').join(kt_annotations.key_by('variantId'))
pprint(kt_rare_variants.schema)
'Analysis group': 'analysis_group',
'AC case': 'ac_case',
'AC ctrl': 'ac_ctrl',
'AN case': 'an_case',
'AN ctrl': 'an_ctrl',
'AF case': 'af_case',
'AF ctrl': 'af_ctrl',
'Estimate': 'est',
'SE': 'se',
'P-value': 'p',
'Comment': 'comment',
}
annotation_expressions = [
'variant_id = %s' % get_expr_for_variant_id(),
'contig = %s' % get_expr_for_contig(),
'pos = %s' % get_expr_for_start_pos(),
"xpos = %s" % get_expr_for_xpos(field_prefix="", pos_field="pos"),
]
kt_variant_annotation = kt_variant_annotation.rename(column_map)
kt_variant_annotation = kt_variant_annotation.annotate('v = Variant(v)')
kt_variant_results = kt_variant_results.rename(column_map)
for expression in annotation_expressions:
kt_variant_results = kt_variant_results.annotate(expression)
kt_variants = kt_variant_results.key_by('v').join(kt_variant_annotation.key_by('v'))
kt_variants = kt_variants.drop(['v'])
kt_results_by_cohort = kt_results_by_cohort.rename(column_map)