# TerraBind: Fast and Accurate Binding Affinity Prediction through Coarse Structural Representations

arXiv: https://arxiv.org/abs/2602.07735 · HTML: https://mlsanity.com/paper/2602.07735/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 50 references that are in the map.
- First posted on arXiv 8 Feb 2026.

## Builds on (most cited first, 11 of 11)

- [Conditional Neural Processes](https://mlsanity.com/paper/1807.01613/) (2018)
- [Learning Local Equivariant Representations for Large-Scale Atomistic Dynamics](https://mlsanity.com/paper/2204.05249/) (2022)
- [e3nn: Euclidean Neural Networks](https://mlsanity.com/paper/2207.09453/) (2022)
- [Efficiently Sampling Functions from Gaussian Process Posteriors](https://mlsanity.com/paper/2002.09309/) (2020)
- [Epistemic Neural Networks](https://mlsanity.com/paper/2107.08924/) (2021)
- [Fine-Tuning Language Models via Epistemic Neural Networks](https://mlsanity.com/paper/2211.01568/) (2022)
- [Equilibrium Matching: Generative Modeling with Implicit Energy-Based Models](https://mlsanity.com/paper/2510.02300/) (2025)
- [NeuralPLexer3: Accurate Biomolecular Complex Structure Prediction with Flow Models](https://mlsanity.com/paper/2412.10743/) (2024)
- [Pearl: A Foundation Model for Placing Every Atom in the Right Location](https://mlsanity.com/paper/2510.24670/) (2025)
- [Boltzina: Efficient and Accurate Virtual Screening via Docking-Guided Binding Prediction with Boltz-2](https://mlsanity.com/paper/2508.17555/) (2025)
- [Pretrained Joint Predictions for Scalable Batch Bayesian Optimization of Molecular Designs](https://mlsanity.com/paper/2511.10590/) (2025)

## Abstract (arXiv)

We present TerraBind, a foundation model for protein-ligand structure and binding affinity prediction that achieves 26-fold faster inference than state-of-the-art methods while improving affinity prediction accuracy by ~20%. Current deep learning approaches to structure-based drug design rely on expensive all-atom diffusion to generate 3D coordinates, creating inference bottlenecks that render large-scale compound screening computationally intractable. We challenge this paradigm with a critical hypothesis: full all-atom resolution is unnecessary for accurate small molecule pose and binding affinity prediction. TerraBind tests this hypothesis through a coarse pocket-level representation (protein Cβ atoms and ligand heavy atoms only) within a multimodal architecture combining COATI-3 molecular encodings and ESM-2 protein embeddings that learns rich structural representations, which are used in a diffusion-free optimization module for pose generation and a binding affinity likelihood prediction module. On structure prediction benchmarks (FoldBench, PoseBusters, Runs N' Poses), TerraBind matches diffusion-based baselines in ligand pose accuracy. Crucially, TerraBind outperforms Boltz-2 by ~20% in Pearson correlation for binding affinity prediction on both a public benchmark (CASP16) and a diverse proprietary dataset (18 biochemical/cell assays). We show that the affinity prediction module also provides well-calibrated affinity uncertainty estimates, addressing a critical gap in reliable compound prioritization for drug discovery. Furthermore, this module enables a continual learning framework and a hedged batch selection strategy that, in simulated drug discovery cycles, achieves 6× greater affinity improvement of selected molecules over greedy-based approaches.