| #!/bin/bash |
| set -euo pipefail |
|
|
| ENCODER="${ENCODER:?must be set}" |
| REFERENCE_PARQUET="${REFERENCE_PARQUET:?must be set}" |
| READS="${READS:?must be set}" |
| REPO_URL="https://github.com/vedatonuryilmaz/dev_genopix.git" |
| REPO_BRANCH="${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 |
|
|
| MODE=$([ "${RESUME:-0}" = "1" ] && echo resume || echo run) |
| QUANT_MODE="${QUANT_MODE:-snk_em}" |
| OUTPUT_NAME="${1:-$(uv run python scripts/run_names.py quantify --encoder "$ENCODER" --reads "$READS" --mode "$QUANT_MODE")}" |
|
|
| if [ "${RESUME:-0}" = "1" ]; then |
| uv run python scripts/hf_artifacts.py pull --repo "$ARTIFACT_REPO" --preset "$OUTPUT_NAME" --dest data/output |
| fi |
|
|
| uv run deepgenopix quantify \ |
| --encoder "$ENCODER" \ |
| --reference-parquet "$REFERENCE_PARQUET" \ |
| --reads "$READS" \ |
| --output-dir "data/output/$OUTPUT_NAME" \ |
| --mode "$QUANT_MODE" \ |
| --run-mode "$MODE" \ |
| --json |
|
|
| uv run python scripts/hf_artifacts.py push --repo "$ARTIFACT_REPO" --preset "$OUTPUT_NAME" --source data/output |
| uv run python scripts/hf_artifacts.py push-quant \ |
| --repo "$DATA_DATASET" \ |
| --output-name "$OUTPUT_NAME" \ |
| --quant-run-dir "data/output/$OUTPUT_NAME" \ |
| --message "quant artifacts from $OUTPUT_NAME" |
|
|
