2025/05/27 by Divya Nori, Nori, Divya, Anisha Parsan +5 · 1 voice · 1 citation
Computer Science · Engineering · Materials Science · #Additive Manufacturing and 3D Printing Technologies #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning in Materials Science #cs.LG
paper · pdf · doi:10.48550/arxiv.2505.21241
openalex publication_date 2025/05/27 · arxiv published 2025/05/27 · arxiv updated 2025/05/27 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Protein binder design has been transformed by hallucination-based methods that optimize structure prediction confidence metrics, such as the interface predicted TM-score (ipTM), via backpropagation. However, these metrics do not reflect the statistical likelihood of a binder-target complex under the learned distribution and yield sparse gradients for optimization. In this work, we propose a method to extract such likelihoods from structure predictors by reinterpreting their confidence outputs as an energy-based model (EBM). By leveraging the Joint Energy-based Modeling (JEM) framework, we introduce pTMEnergy, a statistical energy function derived from predicted inter-residue error distributions. We incorporate pTMEnergy into BindEnergyCraft (BECraft), a design pipeline that maintains the same optimization framework as BindCraft but replaces ipTM with our energy-based objective. BECraft outperforms BindCraft, RFDiffusion, and ESM3 across multiple challenging targets, achieving higher in silico binder success rates while reducing structural clashes. Furthermore, pTMEnergy establishes a new state-of-the-art in structure-based virtual screening tasks for miniprotein and RNA aptamer binders.