SculptDrug : A Spatial Condition-Aware Bayesian Flow Model for Structure-based Drug Design

2025-11-16

AI Summary

Existing structure-based drug design (SBDD) generative models suffer from limitations in modeling binding-site geometric constraints, integrating hierarchical molecular structures, and preserving 3D spatial fidelity. To address these challenges, this work introduces the first Bayesian flow network model that jointly incorporates protein surface geometry and multi-level ligand–protein interactions. Methodologically, we propose a boundary-aware module and a hierarchical encoder, integrated with a progressive denoising mechanism to simultaneously model global binding-pocket constraints and local ligand–protein interactions in atomic-resolution 3D space. This design enables joint optimization of spatial geometric compatibility and structural consistency during generation—achieved for the first time in SBDD. Evaluated on the CrossDocked dataset, our model achieves statistically significant improvements over state-of-the-art methods across three core metrics: spatial alignment accuracy, local contact plausibility, and overall conformational quality.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Search and Optimization: Sampling/Simulation-based Search
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web search models and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Structure-Based drug design (SBDD) has emerged as a popular approach in drug discovery, leveraging three-dimensional protein structures to generate drug ligands. However, existing generative models encounter several key challenges: (1) incorporating boundary condition constraints, (2) integrating hierarchical structural conditions, and (3) ensuring spatial modeling fidelity. To address these limitations, we propose SculptDrug, a spatial condition-aware generative model based on Bayesian flow networks (BFNs). First, SculptDrug follows a BFN-based framework and employs a progressive denoising strategy to ensure spatial modeling fidelity, iteratively refining atom positions while enhancing local interactions for precise spatial alignment. Second, we introduce a Boundary Awareness Block that incorporates protein surface constraints into the generative process to ensure that generated ligands are geometrically compatible with the target protein. Third, we design a Hierarchical Encoder that captures global structural context while preserving fine-grained molecular interactions, ensuring overall consistency and accurate ligand-protein conformations. We evaluate SculptDrug on the CrossDocked dataset, and experimental results demonstrate that SculptDrug outperforms state-of-the-art baselines, highlighting the effectiveness of spatial condition-aware modeling.

Problem

• Incorporating boundary condition constraints in drug ligand generation
• Integrating hierarchical structural conditions for molecular interactions
• Ensuring spatial modeling fidelity for precise ligand-protein alignment

Innovation

• Bayesian flow network framework for progressive denoising
• Boundary awareness block incorporating protein surface constraints
• Hierarchical encoder capturing global and local interactions