#!/usr/bin/env bash
# filter.sh — keep only regions ≥200 bp AND num_sites ≥2
shopt -s nullglob

for f in dmr_TSC_GFP_vs_prime_diff.bed dmr_TSC_GFP_vs_naive_diff.bed dmr_naive_vs_prime_diff.bed
do
  [[ ! -f $f ]] && { echo "File $f not found, skipping"; continue; }

  tmp="${f}.tmp"

  awk 'NR==1 {print; next}
       {len=$3-$2; if (len>=200 && $6>=2) print}' "$f" > "$tmp"

  mv "$tmp" "$f"
  echo "Filtered $f"
done
