Abstract:The task of deducing three-dimensional molecular configurations from their two-dimensional graph representations holds paramount importance in the fields of computational chemistry and pharmaceutical development. The rapid advancement of machine learning, particularly within the domain of deep generative networks, has revolutionized the precision of predictive modeling in this context. Traditional approaches often adopt a two-step strategy: initially estimating interatomic distances and subsequently refining the spatial molecular structure by solving a distance geometry problem. However, this sequential approach occasionally falls short in accurately capturing the intricacies of local atomic arrangements, thereby compromising the fidelity of the resulting structural models. Addressing these limitations, this research introduces a cutting-edge generative framework named \method{}. This framework is grounded in the principles of diffusion observed in classical non-equilibrium thermodynamics. \method{} views atoms as discrete entities and excels in guiding the reversal of diffusion, transforming a distribution of stochastic noise back into coherent molecular structures through a process akin to a Markov chain. This transformation commences with the initial representation of a molecular graph in an abstract latent space, culminating in the realization of three-dimensional structures via a sophisticated bilevel optimization scheme meticulously tailored to meet the specific requirements of the task. One of the formidable challenges in this modeling endeavor involves preserving roto-translational invariance to ensure that the generated molecular conformations adhere to the laws of physics. Extensive experimental evaluations confirm the efficacy of the proposed \method{} in comparison to state-of-the-art methods.

What problem does this paper attempt to address?

The paper aims to address the problem of inferring the 3D configuration of molecules from their 2D graph representations, which is an important task in the field of computational chemistry and drug development. Traditional methods typically employ a two-step strategy: first estimating the interatomic distances, and then optimizing the molecular structure by solving distance geometry problems. However, this approach may fail to accurately capture the complexity of local atomic arrangements, thus affecting the accuracy of the structural models. To tackle this issue, the paper proposes a novel generative framework called DDGF (Diffusion-Driven Generative Framework). Based on the diffusion principles observed in non-equilibrium thermodynamics, DDGF treats atoms as discrete entities and guides the reversal of diffusion through a process similar to a Markov chain, transforming a random noise distribution into an ordered molecular structure. This process starts from the abstract latent space representation of the molecular graph and realizes the 3D structure through a two-layer optimization scheme tailored for specific tasks. The DDGF framework pays particular attention to preserving rotational and translational invariance to ensure that the generated molecular conformations obey the laws of physics. By fine-tuning the weighted variational lower bound and considering the conditional probabilities of conformations, this framework comprehensively deals with the complexity from the beginning to the end of the training. Experimental results demonstrate that DDGF exhibits higher efficiency and accuracy in molecular conformation prediction compared to existing state-of-the-art methods.

Diffusion-Driven Generative Framework for Molecular Conformation Prediction

A Unified Conditional Diffusion Framework for Dual Protein Targets Based Bioactive Molecule Generation

Learning Neural Generative Dynamics for Molecular Conformation Generation

GeoDiff: a Geometric Diffusion Model for Molecular Conformation Generation

Molecular Conformation Generation via Shifting Scores

Prediction of Molecular Conformation using Deep Generative Neural Networks

Geometric-Facilitated Denoising Diffusion Model for 3D Molecule Generation

Equivariant score-based generative diffusion framework for 3D molecules

Torsional Diffusion for Molecular Conformer Generation

Advances in Deep Learning-Based 3D Molecular Generative Models

DiffMD: A Geometric Diffusion Model for Molecular Dynamics Simulations

Modeling Molecular Structures with Intrinsic Diffusion Models

Learning Gradient Fields for Molecular Conformation Generation

Bridging the Gap between Learning and Inference for Diffusion-Based Molecule Generation

Swallowing the Bitter Pill: Simplified Scalable Conformer Generation

3M-Diffusion: Latent Multi-Modal Diffusion for Language-Guided Molecular Structure Generation

SDEGen: learning to evolve molecular conformations from thermodynamic noise for conformation generation

Navigating the Design Space of Equivariant Diffusion-Based Generative Models for De Novo 3D Molecule Generation

Direct Molecular Conformation Generation

DiffSeqMol: A Non-Autoregressive Diffusion-Based Approach for Molecular Sequence Generation and Optimization