#!/usr/bin/env bash
# Oracle VCFs for phase P6 on the 36 GSA oracle GTCs (needs make_oracle_gtcs.sh and fetch_data.sh p6):
#  * gtc2vcf (MIT) → data/p6/oracle/gsa36_gtc2vcf.vcf.gz (one multi-sample VCF, one record per probe)
#  * optional: GTCtoVCF (Apache-2.0) — python2-era code run under python3 with two local compatibility patches,
#    see docs/validation/P6.md; enable with GTCTOVCF=1.
set -euo pipefail
cd "$(dirname "$0")/.."
B=refs/tools/bcftools-1.22
export BCFTOOLS_PLUGINS=$B/plugins
D=data/p0/gsa_v4_pgx
OUT=data/p6/oracle
mkdir -p $OUT data/p6/work
$B/bcftools +gtc2vcf --no-version -t GT,BAF,LRR \
  -b $D/GSA-PGx-48v4-0_20079540_E2_bpm/GSA-PGx-48v4-0_20079540_E2.bpm \
  -c $D/GSA-PGx-48v4-0_20079540_E2_csv/GSA-PGx-48v4-0_20079540_E2.csv \
  -e $D/GSA-PGx-48v4-0_20079540_E2_ClusterFile/GSA-PGx-48v4-0_20079540_E2_ClusterFile.egt \
  -f data/p6/GRCh38.primary.fa -g data/p0/oracle/gsa_gt3_nofma -O u 2> $OUT/gtc2vcf.log \
  | $B/bcftools sort -Oz -o $OUT/gsa36_gtc2vcf.vcf.gz -T data/p6/work/sorttmp
$B/bcftools index -f -t $OUT/gsa36_gtc2vcf.vcf.gz
echo "wrote $OUT/gsa36_gtc2vcf.vcf.gz"