scripts/01_data/summarize_deepstarr.py

#!/usr/bin/env python3
import argparse
import json
from pathlib import Path

import numpy as np
import pandas as pd


def label_values(value):
    if hasattr(value, "tolist"):
        value = value.tolist()
    return [float(value[0]), float(value[1])]


def summarize_split(path: Path):
    df = pd.read_parquet(path)
    seq = df["sequence"].astype(str)
    labels = np.asarray([label_values(v) for v in df["label"]], dtype=float)
    return {
        "rows": int(len(df)),
        "sequence_length_min": int(seq.str.len().min()),
        "sequence_length_median": float(seq.str.len().median()),
        "sequence_length_max": int(seq.str.len().max()),
        "label_0_mean": float(labels[:, 0].mean()),
        "label_1_mean": float(labels[:, 1].mean()),
        "label_sum_mean": float(labels.sum(axis=1).mean()),
    }


def main():
    parser = argparse.ArgumentParser(description="Summarize DeepSTARR parquet splits.")
    parser.add_argument("--dataset_dir", required=True)
    parser.add_argument("--output_json", required=True)
    args = parser.parse_args()

    dataset_dir = Path(args.dataset_dir)
    summary = {}
    for split in ["train", "valid", "test"]:
        path = dataset_dir / f"{split}.parquet"
        if path.exists():
            summary[split] = summarize_split(path)

    output_path = Path(args.output_json)
    output_path.parent.mkdir(parents=True, exist_ok=True)
    output_path.write_text(json.dumps(summary, indent=2), encoding="utf-8")
    print(output_path)


if __name__ == "__main__":
    main()