#!/bin/bash
set -euo pipefail

DAR="/BRC/yan/heart/snACAT_CM/DAR_results/snACAT_CM_DAR_female_vs_male_RUVr_k6.bed"
AR="/BRC/yan/heart/AR_related/masterlist_DHSs_myResults_nonovl_any_chunkIDs.bed"
OUTDIR="/BRC/yan/heart/snACAT_CM/DAR_results"

OUT="${OUTDIR}/snACAT_CM_DAR_RUVr_k6_AR_annotated.bed"

echo "[$(date)] Annotating AR-related peaks..."

# get overlapping peak names
bedtools intersect -a ${DAR} -b ${AR} -u \
  | awk '{print $4}' | sort > /tmp/ar_overlap_ids.txt

echo "  AR-related overlapping peaks: $(wc -l < /tmp/ar_overlap_ids.txt)"

# annotate: add AR_related column
awk 'BEGIN{OFS="\t"}
    NR==FNR { ar[$1]=1; next }
    { print $0, ($4 in ar ? "AR-related" : "non-AR") }
' /tmp/ar_overlap_ids.txt ${DAR} > ${OUT}

echo "  Total peaks annotated: $(wc -l < ${OUT})"
echo "  Output: ${OUT}"
echo "[$(date)] Done."
