Six-Stage Filtering Exposes Weak Drug Generators
HEDGEHOG runs generated molecules through medicinal-chemistry, synthesis, docking, and 3D pose checks, leaving only 0.65% of 230,000 candidates.
Jul 28, 20265 min2607.13155 Code available
Design tools, software metrics, testing, debugging, and programming environments.
HEDGEHOG runs generated molecules through medicinal-chemistry, synthesis, docking, and 3D pose checks, leaving only 0.65% of 230,000 candidates.
Instruction-level binary features train a classifier that recovers 91.75% of unsafe regions at a 6.16% false positive rate.
A consensus protocol achieves optimal resilience while allowing participants to hold different trust assumptions.