| """Download and prepare the filtered K562 and RPE1 datasets used by CausalBench.""" |
|
|
| import argparse |
| from pathlib import Path |
| import subprocess |
|
|
| import numpy as np |
| import pandas as pd |
|
|
| from upstream.preprocessing import preprocess_dataset |
|
|
| ROOT = Path(__file__).resolve().parent |
| RAW = ROOT / "data/raw" |
| DATASETS = { |
| "k562": (35773219, "TabB_K562_day6_summary_stat"), |
| "rpe1": (35775606, "TabC_RPE1_summary_statistic"), |
| } |
| SUMMARY_URL = "https://ars.els-cdn.com/content/image/1-s2.0-S0092867422005979-mmc2.xlsx" |
|
|
|
|
| def download(url, path): |
| if path.exists(): |
| return path |
| path.parent.mkdir(parents=True, exist_ok=True) |
| partial = path.with_suffix(path.suffix + ".part") |
| subprocess.run( |
| ["curl", "--fail", "--location", "--retry", "3", "--continue-at", "-", |
| "--output", str(partial), url], |
| check=True, |
| ) |
| partial.replace(path) |
| return path |
|
|
|
|
| def prepare(name, output_dir): |
| npz = output_dir / f"dataset_{name}_filtered.npz" |
| tsv = output_dir / f"causalbench_{name}.tsv.gz" |
| if npz.exists() and tsv.exists(): |
| print(f"{name}: using existing files in {output_dir}") |
| return |
| if npz.exists() or tsv.exists(): |
| raise FileExistsError("Output is incomplete. Choose a new --output-dir to rebuild.") |
|
|
| _, sheet = DATASETS[name] |
| summary = pd.read_excel(RAW / "summary_stats.xlsx", sheet_name=sheet) |
| expression, genes, labels = preprocess_dataset(str(RAW / f"{name}.h5ad"), summary) |
| output_dir.mkdir(parents=True, exist_ok=True) |
| np.savez(npz, expression_matrix=expression, var_names=np.asarray(genes, dtype=str), |
| interventions=np.asarray(labels, dtype=str)) |
| table = pd.DataFrame(expression, columns=genes) |
| table["intervention"] = labels |
| table.to_csv(tsv, sep="\t", index=False, compression="gzip") |
| print(f"{name}: saved {expression.shape[0]:,} cells × {expression.shape[1]} genes") |
|
|
|
|
| if __name__ == "__main__": |
| parser = argparse.ArgumentParser(description=__doc__) |
| parser.add_argument("dataset", choices=["k562", "rpe1", "all"], nargs="?", default="all") |
| parser.add_argument("--download-only", action="store_true") |
| parser.add_argument("--output-dir", type=Path, default=ROOT / "data/prepared") |
| args = parser.parse_args() |
| names = DATASETS if args.dataset == "all" else [args.dataset] |
| download(SUMMARY_URL, RAW / "summary_stats.xlsx") |
| for name in names: |
| file_id, _ = DATASETS[name] |
| download(f"https://ndownloader.figshare.com/files/{file_id}", RAW / f"{name}.h5ad") |
| if args.download_only: |
| print(f"{name}: raw data ready in {RAW}") |
| else: |
| prepare(name, args.output_dir) |
|
|
