#!/bin/bash -Ceuo pipefail
SHARD1="chr1,chr4,chr15,chr19,chr22,chrY,chr2,chr8,chr9,chr17,chr18,chr20"
SHARD2="chr3,chr5,chr10,chr11,chr13,chr14,chr6,chr7,chr12,chr16,chr21,chrX"

allele_counts -t 2 -r "$SHARD1" 1029_0W2-T1-TDNA-1_A23KCGYLT4_1.recalibrated.bam dbsnp138.snps.hg38.tsv.gz \
    "sample_id=1029_0W2-T1-TDNA-1_A23KCGYLT4_1" "specimen_id=1029_0W2" \
    "flowcell=unknown" "sample_type=tumor" > allele_s1.tsv &
allele_counts -t 2 -r "$SHARD2" 1029_0W2-T1-TDNA-1_A23KCGYLT4_1.recalibrated.bam dbsnp138.snps.hg38.tsv.gz \
    "sample_id=1029_0W2-T1-TDNA-1_A23KCGYLT4_1" "specimen_id=1029_0W2" \
    "flowcell=unknown" "sample_type=tumor" > allele_s2.tsv &
wait

{ head -1 allele_s1.tsv | cut -f2-;
  for f in allele_s*.tsv; do tail -n +2 "$f"; done | sort -t'	' -k1,1n | cut -f2-;
} | gzip > 1029_0W2-T1-TDNA-1_A23KCGYLT4_1_alleles.tsv.gz

cat <<-END_VERSIONS > versions.yml
"NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_SOMATIC_ALL:SOMATIC_CNV_CALLING:SOMA_CNV_ALLELE_COUNTS":
    soma-cnv: 20260604-fbf83cd
END_VERSIONS