File size: 2,477 Bytes
1bc9257 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 | """Illustrative study/content-grouped split, not an official benchmark protocol."""
from __future__ import annotations
import argparse
import hashlib
from pathlib import Path
import pandas as pd
from hesrt_io import RELEASE_TAG, load_catalog
def assign_groups(frame: pd.DataFrame, seed: str) -> pd.DataFrame:
"""Keep each GSE and any exact image/expression duplicates in one component."""
frame = frame.reset_index(drop=True)
parent = list(range(len(frame)))
def root(i: int) -> int:
while parent[i] != i:
parent[i] = parent[parent[i]]
i = parent[i]
return i
for column in ["gse_id", "image_sha256", "expression_sha256"]:
first: dict[str, int] = {}
for i, value in enumerate(frame[column]):
if pd.isna(value) or not str(value).strip():
continue
value = str(value)
if value in first:
parent[root(i)] = root(first[value])
else:
first[value] = i
labels: dict[int, str] = {}
for i, gsm in enumerate(frame.gsm_id):
r = root(i)
labels[r] = min(labels.get(r, str(gsm)), str(gsm))
result = frame[["gse_id", "gsm_id"]].copy()
result["leakage_group"] = [labels[root(i)] for i in range(len(frame))]
def split(group: str) -> str:
fraction = int(hashlib.sha256((seed + ":" + group).encode()).hexdigest()[:16], 16) / 2**64
return "train" if fraction < 0.8 else "validation" if fraction < 0.9 else "test"
result["split"] = result.leakage_group.map(split)
return result
def main() -> None:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--revision", default=RELEASE_TAG)
parser.add_argument("--seed", default="hesrt-example-0")
parser.add_argument("--output", type=Path, default=Path("study_split.csv"))
args = parser.parse_args()
frame, revision = load_catalog(args.revision)
result = assign_groups(frame, args.seed)
result["revision"] = revision
args.output.parent.mkdir(parents=True, exist_ok=True)
result.to_csv(args.output, index=False)
print(result.groupby("split").agg(samples=("gsm_id", "size"), studies=("gse_id", "nunique")))
print("Approximate 80/10/10 group allocation; not stratified and not donor-disjoint certification.")
print("Resolve shared donors, near-duplicates, cohort overlap and task eligibility separately.")
if __name__ == "__main__":
main()
|