#!/usr/bin/env bash
# Usage: simulate_query_one.sh genome.fna.gz output_dir
# Simulates a fixed-size paired-end HiSeq read set for the query benchmark.
# Unlike simulate_one.sh (coverage-proportional, used to build the indexed
# specimens), this always draws N_READS pairs regardless of genome size —
# query benchmark numbers (wall/RSS) must stay comparable across genomes of
# very different sizes. Independent iss run (unseeded), so error draw
# differs from any simulated_data/ reads for the same genome.
set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
ISS="${SCRIPT_DIR}/../.venv/bin/iss"
N_READS=100000
CPUS="${CPUS:-$(sysctl -n hw.logicalcpu 2>/dev/null || nproc 2>/dev/null || echo 2)}"

genome_file="$1"
out_dir="$2"

mkdir -p "${out_dir}"

tmp_fasta=$(mktemp "${TMPDIR:-/tmp}/obikmer_XXXXXX.fna")
trap 'rm -f "${tmp_fasta}"' EXIT

gzip -dc "${genome_file}" > "${tmp_fasta}"

echo "[${out_dir}]  ${N_READS} read pairs (query benchmark, fixed size)"

"${ISS}" generate \
    --genomes   "${tmp_fasta}" \
    --model     HiSeq \
    --n_reads   "${N_READS}" \
    --cpus      "${CPUS}" \
    --compress \
    --output    "${out_dir}/reads"