---
title: "UniPocket: Physics-Aware Geometric Graph Learning with Manifold Completeness for Ligand-Specific Binding Site Prediction"
authors:
  - "Kangxin Chen"
  - "Jieyu Zhao"
  - "Jinli Hu"
  - "Min Xie"
doi: "10.24963/ijcai.2026/610"
url: "https://ijcai.org/proceedings/2026/610"
pdf: "https://ijcai.org/proceedings/2026/0610.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "5477-5485"
track: "Main Track"
keywords:
  - "Multidisciplinary Topics and Applications: Bioinformatics"
---

# UniPocket: Physics-Aware Geometric Graph Learning with Manifold Completeness for Ligand-Specific Binding Site Prediction

Kangxin Chen, Jieyu Zhao, Jinli Hu, Min Xie

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 5477-5485. DOI: [10.24963/ijcai.2026/610](https://doi.org/10.24963/ijcai.2026/610)

## Abstract

Predicting ligand binding sites on protein surfaces requires capturing complex local geometries and satisfying physical constraints. Existing voxel-based methods suffer from high computational costs and rotation sensitivity, while standard point-cloud GNNs often lack geometric completeness—failing to distinguish chiral structures or subtle topological variations. In this work, we propose UniPocket, a novel E(3)-equivariant surface graph neural network. Inspired by recent advances in efficient geometric completeness, UniPocket constructs a Manifold-Aware Surface Encoder that utilizes local surface normals as virtual reference frames to capture complete geometric invariants without expensive high-order tensor products. Furthermore, we introduce a Ligand-Gated Message Passing mechanism to condition the surface features on the chemical semantics of the target ligand, and a Physics-Aware Vector Rejection Module that enforces steric constraints via orthogonal vector decomposition. Experimental results on standard benchmarks (PDBbind, COACH420, Holo4k) demonstrate that UniPocket achieves state-of-the-art performance, validating that geometric completeness and physical inductive biases are key to precise binding site detection.