Abstract:Navigating the vast chemical space of druggable compounds is a formidable challenge in drug discovery, where generative models are increasingly employed to identify viable candidates. Conditional 3D structure-based drug design (3D-SBDD) models, which take into account complex three-dimensional interactions and molecular geometries, are particularly promising. Scaffold hopping is an efficient strategy that facilitates the identification of similar active compounds by strategically modifying the core structure of molecules, effectively narrowing the wide chemical space and enhancing the discovery of drug-like products. However, the practical application of 3D-SBDD generative models is hampered by their slow processing speeds. To address this bottleneck, we introduce TurboHopp, an accelerated pocket-conditioned 3D scaffold hopping model that merges the strategic effectiveness of traditional scaffold hopping with rapid generation capabilities of consistency models. This synergy not only enhances efficiency but also significantly boosts generation speeds, achieving up to 30 times faster inference speed as well as superior generation quality compared to existing diffusion-based models, establishing TurboHopp as a powerful tool in drug discovery. Supported by faster inference speed, we further optimize our model, using Reinforcement Learning for Consistency Models (RLCM), to output desirable molecules. We demonstrate the broad applicability of TurboHopp across multiple drug discovery scenarios, underscoring its potential in diverse molecular settings.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is the slow processing speed of the drug design model under 3D structure conditions (3D - SBDD) in the drug discovery process. Specifically: 1. **Accelerating molecular scaffold hopping**: Traditional 3D - SBDD generation models have very slow inference speeds due to their iterative sampling processes, which greatly limit their efficiency and practicality in real - world applications. To solve this problem, the paper introduces the TurboHopp model, an E(3) - equivariant consistency model that combines fast - generation capabilities with traditional scaffold - hopping strategies. 2. **Improving generation quality**: In addition to the speed improvement, TurboHopp also aims to generate molecules with higher pharmacodynamic properties, synthesizability, connectivity, and binding affinity. By introducing Reinforcement Learning for Consistency Models (RLCM), the binding affinity of generated molecules is further optimized and steric clashes are reduced without the need for redocking. 3. **Enabling real - time applications**: Faster inference speeds make real - time applications possible, such as providing instant feedback and facilitating iterative testing during interactive modeling sessions with human experts. ### Specific problem summary - **Slow inference**: Existing 3D - SBDD generation models require hundreds to thousands of iterations to complete one inference, resulting in extremely slow processing speeds. - **Insufficient generation quality**: The molecular conformations generated by existing models often do not meet key biophysical constraints, such as high strain energy and protein - collision problems. - **Difficulty in optimizing specific targets**: Direct modeling is difficult to optimize certain complex targets, such as binding affinity and avoiding steric clashes. ### Solutions - **TurboHopp model**: By combining the fast - generation capabilities of consistency models and the effective strategies of scaffold hopping, the generation speed is significantly improved (up to 30 times). - **RLCM optimization**: Utilize reinforcement learning techniques to further optimize specific properties of generated molecules, such as binding affinity and avoiding steric clashes. - **Flexible multi - step generation**: Adopt a metric - based sampling method to dynamically adjust the generation process to ensure optimal results. Through these improvements, TurboHopp not only improves the generation speed but also shows broad application potential in multiple drug discovery scenarios.

TurboHopp: Accelerated Molecule Scaffold Hopping with Consistency Models

DiffHopp: A Graph Diffusion Model for Novel Drug Design via Scaffold Hopping

Scaffold-Constrained Molecular Generation

A scaffold‐based deep generative model considering molecular stereochemical information

Conditional Latent Space Molecular Scaffold Optimization for Accelerated Molecular Design

3D-SMGE: a pipeline for scaffold-based molecular generation and evaluation

Sc2Mol: a Scaffold-Based Two-Step Molecule Generator with Variational Autoencoder and Transformer.

Learning to Extend Molecular Scaffolds with Structural Motifs

Scaffold-Based Multi-Objective Drug Candidate Optimization

DecompOpt: Controllable and Decomposed Diffusion Models for Structure-based Molecular Optimization

Pocket Crafter: a 3D generative modeling based workflow for the rapid generation of hit molecules in drug discovery

DrugHIVE: Target-specific spatial drug design and optimization with a hierarchical generative model

MolSnapper: Conditioning Diffusion for Structure Based Drug Design

3D-Scaffold: A Deep Learning Framework to Generate 3D Coordinates of Drug-like Molecules with Desired Scaffolds

Scaffold-based molecular design using graph generative model

Molecular generation strategy and optimization based on A2C reinforcement learning in de novo drug design

ScaffoldGVAE: scaffold generation and hopping of drug molecules via a variational autoencoder based on multi-view graph neural networks

Structure-Based Molecule Optimization via Gradient-Guided Bayesian Update

SMILES-based deep generative scaffold decorator for de-novo drug design

SynthFormer: Equivariant Pharmacophore-based Generation of Molecules for Ligand-Based Drug Design

A dual diffusion model enables 3D molecule generation and lead optimization based on target pockets