Datasets:
File size: 3,097 Bytes
84a242a | 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 64 65 66 67 | #!/usr/bin/env python3
"""Reconstruct native MolAS evaluation labels from unchanged raw tables.
This is deterministic data preparation, not docking or model training. The
output is explicitly reconstructed; it is not the missing original server CSV.
Acquisition uses a different, continuous target and is never overwritten here.
"""
import argparse
import csv
import hashlib
import math
from pathlib import Path
ALGORITHMS = ['surfdock', 'unimol', 'diffdock', 'diffdock_l', 'gnina', 'autodock', 'smina']
def native_score(rmsd, pb_ratio):
if rmsd > 5 or rmsd < 0:
return 0.0
return (1 + math.exp(-6)) / (1 + math.exp(3 * (rmsd - 2))) * (pb_ratio == 1.0)
def read(path):
with path.open(newline='') as handle:
return list(csv.DictReader(handle))
def main():
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument('--root', type=Path, default=Path(__file__).resolve().parents[1])
parser.add_argument('--output', type=Path, help='Explicit reconstructed target destination')
parser.add_argument('--check-only', action='store_true', help='Compare with the existing reconstructed file without writing')
args = parser.parse_args()
root = args.root.resolve()
output = args.output or root / 'data/targets/moad_score_native_reconstructed.csv'
protected = [root / f'data/raw/moad_{name}_trimmed.csv' for name in ['rmsd', 'pb_ratio', 'runtime']]
protected += [root / 'data/targets/strict_score_matrix.csv', root / 'data/targets/full_strict_score_matrix.csv']
if output.resolve() in {path.resolve() for path in protected}:
parser.error('Refusing to overwrite raw data or an original acquisition target')
rmsd = read(root / 'data/raw/moad_rmsd_trimmed.csv')
pb = read(root / 'data/raw/moad_pb_ratio_trimmed.csv')
ids = [row['PDB_CCD_ID'] for row in rmsd]
assert len(ids) == len(set(ids)) == 3179
assert ids == [row['PDB_CCD_ID'] for row in pb]
records = [[identifier] + [native_score(float(r[a + '_rmsd']), float(p[a + '_pb_ratio']))
for a in ALGORITHMS]
for identifier, r, p in zip(ids, rmsd, pb)]
if args.check_only:
existing = read(output)
assert [row['PDB_CCD_ID'] for row in existing] == ids
largest = max(abs(float(row[a]) - record[j + 1])
for row, record in zip(existing, records) for j, a in enumerate(ALGORITHMS))
assert largest <= 1e-12, f'Native target mismatch: {largest}'
print(f'Checked 3179 x 7 native labels; maximum absolute formula difference: {largest}')
else:
output.parent.mkdir(parents=True, exist_ok=True)
with output.open('w', newline='') as handle:
writer = csv.writer(handle)
writer.writerow(['PDB_CCD_ID'] + ALGORITHMS)
writer.writerows([row[0]] + [repr(x) for x in row[1:]] for row in records)
print(f'Wrote explicitly reconstructed native evaluation target: {output}')
print(f'SHA256: {hashlib.sha256(output.read_bytes()).hexdigest()}')
if __name__ == '__main__':
main()
|