| |
| |
| |
|
|
| """Reorder MotifBench's case summary before running its group-summary script.""" |
| import argparse |
| import csv |
| import importlib.util |
| from pathlib import Path |
|
|
|
|
| def main(): |
| parser = argparse.ArgumentParser(description=__doc__) |
| parser.add_argument("work", type=Path, help="prepared MotifBench working directory") |
| parser.add_argument("results", type=Path, help="evaluation output directory") |
| args = parser.parse_args() |
| with (args.work / "test_cases.csv").open(newline="") as handle: |
| cases = list(csv.DictReader(handle)) |
| order = {f"{i:02d}_{row['pdb_id']}": i for i, row in enumerate(cases, 1)} |
| with (args.results / "summary_by_problem.csv").open(newline="") as handle: |
| reader = csv.DictReader(handle) |
| fields, rows = reader.fieldnames, list(reader) |
| if not rows: |
| raise ValueError("the upstream summary contains no cases") |
| names = [row["Problem"] for row in rows] |
| if len(set(names)) != len(names) or any(name not in order for name in names): |
| raise ValueError("summary contains duplicate or unknown case names") |
| for row in rows: |
| if int(row["Num_Evaluated"]) != 100: |
| raise ValueError(f"{row['Problem']}: upstream score assumes 100 evaluated designs") |
| cluster = args.results / row["Problem"] / "esm_successful_backbones/diversity_cluster.tsv" |
| if int(row["Num_Solutions"]) > 0 and not cluster.is_file(): |
| raise FileNotFoundError(f"missing diversity clusters: {cluster}") |
| ordered = args.results / "summary_by_problem_ordered.csv" |
| overall = args.results / "overall_summary_ordered.csv" |
| if ordered.exists() or overall.exists(): |
| raise FileExistsError("ordered summary output already exists") |
| rows.sort(key=lambda row: order[row["Problem"]]) |
| with ordered.open("x", newline="") as handle: |
| writer = csv.DictWriter(handle, fieldnames=fields) |
| writer.writeheader() |
| writer.writerows(rows) |
| source = args.work / "scripts/write_overall_summary.py" |
| module_spec = importlib.util.spec_from_file_location("motifbench_summary", source) |
| if module_spec is None or module_spec.loader is None: |
| raise ImportError(f"cannot load MotifBench summary module: {source}") |
| module = importlib.util.module_from_spec(module_spec) |
| module_spec.loader.exec_module(module) |
| module.np.random.seed(0) |
| module.main(str(args.work / "test_cases.csv"), str(ordered), str(overall)) |
| module.np.random.seed(0) |
| for row in rows: |
| cluster = module.diversity_cluster_path(str(args.results), row["Problem"]) |
| successes = module.load_successes_and_cluster(cluster) |
| row["overall_score"] = module.MBScore_boostrap(successes)[0] |
| with ordered.open("w", newline="") as handle: |
| writer = csv.DictWriter(handle, fieldnames=fields + ["overall_score"]) |
| writer.writeheader() |
| writer.writerows(rows) |
| print(f"Summarized {len(rows)}/{len(cases)} cases in {ordered} and {overall}.") |
| if len(rows) != len(cases): |
| print("Warning: this is a partial benchmark summary, not a full BILBY-PRO score.") |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
