#!/bin/bash

QUERY=/BRC/yan/heart/amina_res/DAR_CM_Female_YvsO_ruvk4_FDR001_logFC58.bed
DAR_ER=/BRC/yan/heart/hormon/filtered_bed/DAR_ER.bed
ER_PEAKS=/BRC/yan/heart/ER_related/ER_peaks.bed
OUTDIR=/BRC/yan/heart/amina_res/overlap_results
SUMMARY=$OUTDIR/overlap_summary.tsv

mkdir -p "$OUTDIR"

QUERY_N=$(wc -l < "$QUERY")
DAR_ER_N=$(wc -l < "$DAR_ER")
ER_PEAKS_N=$(wc -l < "$ER_PEAKS")

# Overlap 1: DAR vs DAR_ER
bedtools intersect -a "$QUERY" -b "$DAR_ER" -u \
    > "$OUTDIR/DAR_overlap_DAR_ER.bed"
OVL1=$(wc -l < "$OUTDIR/DAR_overlap_DAR_ER.bed")

# Overlap 2: DAR vs ER_peaks
bedtools intersect -a "$QUERY" -b "$ER_PEAKS" -u \
    > "$OUTDIR/DAR_overlap_ER_peaks.bed"
OVL2=$(wc -l < "$OUTDIR/DAR_overlap_ER_peaks.bed")

# Summary table
{
printf "%-30s\t%s\t%s\t%s\t%s\n" \
    "Comparison" "Query_regions" "Reference_regions" "Overlapping_query" "Pct_query_overlapping"
printf "%-30s\t%d\t%d\t%d\t%.2f%%\n" \
    "DAR_CM_Female vs DAR_ER" \
    "$QUERY_N" "$DAR_ER_N" "$OVL1" \
    "$(echo "scale=4; $OVL1/$QUERY_N*100" | bc)"
printf "%-30s\t%d\t%d\t%d\t%.2f%%\n" \
    "DAR_CM_Female vs ER_peaks" \
    "$QUERY_N" "$ER_PEAKS_N" "$OVL2" \
    "$(echo "scale=4; $OVL2/$QUERY_N*100" | bc)"
} | tee "$SUMMARY"

echo ""
echo "Output files:"
echo "  $OUTDIR/DAR_overlap_DAR_ER.bed"
echo "  $OUTDIR/DAR_overlap_ER_peaks.bed"
echo "  $SUMMARY"
