Inference-Time Projection for Physically Valid Biomolecular Diffusion Models
Quick Answer
This paper shows that The authors present a novel inference-time projection method for AlphaFold 3-style models, enhancing physical validity without retraining.
Quick Take
Their approach, applied to Boltz-2 and OpenFold-3 across five benchmarks, achieves perfect physical validity while maintaining structural accuracy and minimal runtime overhead.
Key Points
- Introduces two closed-form projection operators for physical validity in biomolecular models.
- Achieves perfect physical validity with negligible runtime and memory overhead.
- Applicable to any AF3-style sampler without retraining or network evaluations.
- Demonstrated on Boltz-2 and OpenFold-3 across five benchmarks including CASP15 and CASP16.
- Maintains structural accuracy and ligand placement metrics during projection.
Paper Resources
📖 Reader Mode
~2 min readAbstract:AlphaFold 3-style cofolding models predict biomolecular complexes with high structural accuracy, yet a large fraction of their outputs are physically invalid: chains overlap at interfaces, ligand bond lengths and angles are distorted, rings are non-planar, and stereocentres are inverted. Current approaches either steer the sampler with physics-informed potentials, which multiplies sampling cost and memory overhead making inference impossible on large complexes, or finetune the model, costing time and tying the fix to one architecture. We observe that, unlike structural accuracy, physical validity is fully verifiable at inference time from quantities the sampler already holds. We therefore treat physical validity as a constrained inference problem and introduce two closed-form projection operators applied to the diffusion model's denoised clean-coordinate estimate, $\hat{x}_0$: an inter-chain van der Waals projection that pushes apart the most severely clashing atom pairs, and a ligand distance-geometry projection that restores bond lengths, angles, internal contacts, planarity and chirality. Both operators are local, sparse and displacement-capped, require no network evaluations, gradients or importance sampling, and leave the denoiser and its weights untouched, so they can be dropped into any AF3-style sampler without retraining. Applied to two independently developed models, Boltz-2 and OpenFold-3, across five benchmarks (CASP15, CASP16, the PoseBusters monomer and complex sets, and the Boltz physical-validity test set), our method recovers perfect physical validity while preserving structural-accuracy and ligand-placement metrics. These gains are achieved with negligible runtime and memory overhead, providing a practical, model-agnostic route to physically valid all-atom structure prediction.
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.07037 [cs.AI] |
| (or arXiv:2610.07037v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07037 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Q Qurat-Ul-Ain [view email]
[v1]
Sun, 4 Oct 2026 23:15:03 UTC (12,299 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →HOBA: Hierarchical On-Policy Bidding Agents for Adaptive Online Advertising
HOBA (Hierarchical On-policy Bidding Agents) is a novel hierarchical reinforcement learning framework that enhances online advertising bidding systems by improving adaptability and reducing hyperparameter tuning costs. It utilizes a for hyperparameter inference, a SARSA agent for expert model selection, and a dynamic expert pool for bid execution, achieving a +3.6% increase in target cost during large-scale deployment and outperforming state-of-the-art baselines on AuctionNet.