3D molecule generation by denoising voxel grids

Pedro O. Pinheiro,Joshua Rackers,Joseph Kleinhenz,Michael Maser,Omar Mahmood,Andrew Martin Watkins,Stephen Ra,Vishnu Sresht,Saeed Saremi

2024-03-09

Abstract:We propose a new score-based approach to generate 3D molecules represented as atomic densities on regular grids. First, we train a denoising neural network that learns to map from a smooth distribution of noisy molecules to the distribution of real molecules. Then, we follow the neural empirical Bayes framework (Saremi and Hyvarinen, 19) and generate molecules in two steps: (i) sample noisy density grids from a smooth distribution via underdamped Langevin Markov chain Monte Carlo, and (ii) recover the "clean" molecule by denoising the noisy grid with a single step. Our method, VoxMol, generates molecules in a fundamentally different way than the current state of the art (ie, diffusion models applied to atom point clouds). It differs in terms of the data representation, the noise model, the network architecture and the generative modeling algorithm. Our experiments show that VoxMol captures the distribution of drug-like molecules better than state of the art, while being faster to generate samples.

Machine Learning,Quantitative Methods

What problem does this paper attempt to address?

The problem addressed in this paper is 3D molecular generation, that is, how to create a continuous atomic density model representing chemical compounds in three-dimensional space. Existing methods typically rely on diffusion models on point clouds, but these approaches have limitations such as requiring prior knowledge of the number of atoms, handling different types of features (such as continuous and discrete), and difficulties in capturing long-range dependencies. The paper proposes a new approach called VoxMol, which uses denoising voxel grid techniques to generate 3D molecules. VoxMol trains a neural network to sample from a smooth noise distribution and denoise voxelized molecular grids, then generates clean molecular samples using a "walk-jump" sampling strategy. This approach differs from current techniques in data representation, noise modeling, network architecture, and generative modeling algorithms, and performs better in dealing with larger, drug-scale molecules. The paper demonstrates that VoxMol outperforms existing methods in generating drug-like molecules with desired properties.

3D molecule generation by denoising voxel grids

Structure-based drug design by denoising voxel grids

3DMolNet: A Generative Network for Molecular Structures

NEBULA: Neural Empirical Bayes Under Latent Representations for Efficient and Controllable Design of Molecular Libraries

Nc-vae: normalised conditional diverse variational autoencoder guided de novo molecule generation

Geometric-Facilitated Denoising Diffusion Model for 3D Molecule Generation

Lift Your Molecules: Molecular Graph Generation in Latent Euclidean Space

Neural Volumetric Mesh Generator

Learning Neural Generative Dynamics for Molecular Conformation Generation

Unified Generative Modeling of 3D Molecules with Bayesian Flow Networks

Advances in Deep Learning-Based 3D Molecular Generative Models

Equivariant score-based generative diffusion framework for 3D molecules

APMG: 3D Molecule Generation Driven by Atomic Chemical Properties

Zero Shot Molecular Generation via Similarity Kernels

Graph-Based Generative Adversarial Networks for Molecular Generation with Noise Diffusion.

MolecularRNN: Generating realistic molecular graphs with optimized properties

Unified Generative Modeling of 3D Molecules via Bayesian Flow Networks

Von Mises Mixture Distributions for Molecular Conformation Generation

Molecule3D: A Benchmark for Predicting 3D Geometries from Molecular Graphs

MOLUCINATE: A Generative Model for Molecules in 3D Space

Generating 3D Molecular Structures Conditional on a Receptor Binding Site with Deep Generative Models