#!/bin/bash
set -euo pipefail

ENCODER="${ENCODER:?must be set}"
REFERENCE_PARQUET="${REFERENCE_PARQUET:?must be set}"
QUANT_RUN_DIR="${QUANT_RUN_DIR:?must be set}"
QUANT_OUTPUT_NAME="${QUANT_OUTPUT_NAME:?must be set to the remote_quantify.sh output name}"
REPO_URL="https://github.com/vedatonuryilmaz/dev_genopix.git"
REPO_BRANCH="${REPO_BRANCH:-ml-intern}"
DATA_DATASET="${DATA_DATASET:-vedatonuryilmaz/te-seqdata-v1}"
ARTIFACT_REPO="${ARTIFACT_REPO:?must be set}"

git clone --branch "$REPO_BRANCH" --depth 1 "$REPO_URL" deepgenopix
cd deepgenopix
uv sync --extra hf --extra quant

OUTPUT_NAME="${1:-$(uv run python scripts/run_names.py variant --quant-run-name "$QUANT_OUTPUT_NAME" ${NULL_STATS:+--null-stats "$NULL_STATS"})}"

if [ "${RESUME:-0}" = "1" ]; then
  uv run python scripts/hf_artifacts.py pull --repo "$ARTIFACT_REPO" --preset "$OUTPUT_NAME" --dest data/output
fi

uv run python scripts/hf_artifacts.py pull-quant \
  --repo "$DATA_DATASET" \
  --output-name "$QUANT_OUTPUT_NAME" \
  --quant-run-dir "$QUANT_RUN_DIR"

MODE=$([ "${RESUME:-0}" = "1" ] && echo resume || echo run)
uv run deepgenopix variant \
  --encoder "$ENCODER" \
  --reference-parquet "$REFERENCE_PARQUET" \
  --quant-run-dir "$QUANT_RUN_DIR" \
  --output-vcf "data/output/$OUTPUT_NAME/variants.vcf" \
  ${NULL_STATS:+--null-stats "$NULL_STATS"} \
  --run-mode "$MODE" \
  --json

uv run python scripts/hf_artifacts.py push --repo "$ARTIFACT_REPO" --preset "$OUTPUT_NAME" --source data/output