Boltz-Perturb: Probing Generalization in Co-Folding Models via Inference-Time Perturbation

Hyeyun Jung, Alan C Cheng, BoRam Lee

Keywords: deep generative models, diffusion models, protein structure prediction, drug discovery, sampling diversity, perturbation, memorization, generalization, AlphaFold3, co-folding models, training-free, biomolecular modeling

TL;DR: Training-free perturbation of conditioning signals in AlphaFold3-style co-folding models reveals that low-diversity outputs reflect a sampling deficiency rather than a generalization failure, and unlocks correct binding modes without retraining.

Abstract:

AlphaFold3-style co-folding models are powerful deep generative models (DGMs) for biomolecular structure prediction, yet they exhibit low-diversity outputs that frequently miss correct small molecule binding modes.
We present Boltz-Perturb, a training-free perturbation framework that injects time-annealed noise into the conditioning signals of Boltz-2's denoising module during inference. Through true-coordinate injection experiments, we establish that the model's learned energy landscape already encodes correct binding-mode basins, reframing the problem as a sampling deficiency. We introduce Token Bias Perturbation (TBP) and Token Conditioning Perturbation (TCP), targeting attention biases and token embeddings, respectively. Notably, TCP raised oracle success rate from 17.7\% to 30.6\% using three-fold fewer samples across 62 targets. Our results support that co-folding models learn more than they express, but redundant, sample-identical conditioning masks the sampling paths toward correct structures. To our knowledge, this is the first perturbation analysis of an AF3-style co-folding architecture for small-molecule binding mode diversity, demonstrating that conditioning perturbation can unlock latent capacity for drug discovery without retraining.

Submission Number: 213