#!/bin/bash -Ceuo pipefail
# Build tight BED from given sites (point intervals)
# `|| true` guards against grep exit 1 on zero data lines (pipefail-safe)
zcat Sig_18_Blood.stripped.vcf.gz | { grep -v '^#' || true; } | \
    awk 'BEGIN{OFS="\t"}{print $1, $2-1, $2+length($4)-1}' \
    > sites.bed

if [ ! -s sites.bed ]; then
    # No survivors from pre-filter — emit empty VCF with correct header
    zcat Sig_18_Blood.stripped.vcf.gz | { grep '^#' || true; } | bgzip > Sig_18_Blood.tumor_genotyped.vcf.gz
    tabix -p vcf Sig_18_Blood.tumor_genotyped.vcf.gz
else
    sentieon driver \
        -r Homo_sapiens_assembly38.fasta \
        -t 12 \
        --interval sites.bed \
        --interval_padding 200 \
        -i Sig_18_tissue.recalibrated.bam \
        --algo TNhaplotyper2 \
        --tumor_sample Sig_18_tissue \
        --given Sig_18_Blood.stripped.vcf.gz \
         \
        Sig_18_Blood.tumor_genotyped.vcf.gz
fi

cat <<-END_VERSIONS > versions.yml
"NFCORE_SAREK:SAREK:BAM_CHIP_DETECTION:SENTIEON_GENOTYPE":
    sentieon: $(sentieon driver --version 2>&1 | sed -n 's/^sentieon-genomics-//p')
END_VERSIONS