BindEnergyCraft: Casting Protein Structure Predictors as Energy-Based Models for Binder Design

2025-05-27

AI Summary

Current protein binder design relies on structural confidence metrics (e.g., ipTM) for optimization, but these metrics lack statistical interpretability and exhibit sparse gradients. Method: We reinterpret structure predictors (e.g., AlphaFold2, ESM) as energy models and introduce pTMEnergy—a differentiable, probabilistically consistent energy function that explicitly models the statistical likelihood of binding complexes under the learned distribution. Building upon this, we propose BindEnergyCraft: a framework enabling end-to-end sequence–structure co-optimization by substituting ipTM with pTMEnergy, integrated with Joint Energy Modeling (JEM)-inspired energy learning and confidence recalibration. Contribution/Results: On multiple challenging targets, BindEnergyCraft significantly outperforms BindCraft, RFDiffusion, and ESM3—improving *in silico* binding success rates, reducing structural clashes, and achieving new state-of-the-art performance in mini-protein and RNA aptamer virtual screening.

Technology Category

• Machine Learning: Bio-inspired Learning
• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Prompt Engineering / Prompting

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

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.

Problem

• Optimizing protein binder design using energy-based models
• Improving statistical likelihood of binder-target complexes
• Enhancing structure-based virtual screening performance

Innovation

• Reinterpreting confidence outputs as energy-based model
• Introducing pTMEnergy from predicted error distributions
• Optimizing design pipeline with energy-based objective