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.

Efficient Generation of Conformer Ensembles Using Internal Coordinates and a Generative Directional Graph Convolution Neural Network

CoarsenConf: Equivariant Coarsening with Aggregated Attention for Molecular Conformer Generation

Learning Neural Generative Dynamics for Molecular Conformation Generation

Lift Your Molecules: Molecular Graph Generation in Latent Euclidean Space

Prediction of Molecular Conformation using Deep Generative Neural Networks

TorsionNet: A Reinforcement Learning Approach to Sequential Conformer Search

Diffusion-Driven Generative Framework for Molecular Conformation Prediction

Torsional Diffusion for Molecular Conformer Generation

Leveraging 2D molecular graph pretraining for improved 3D conformer generation with graph neural networks

Structure-Aware E(3)-Invariant Molecular Conformer Aggregation Networks

Pre-training of Molecular GNNs via Conditional Boltzmann Generator

Data-Efficient Generation of Protein Conformational Ensembles with Backbone-to-Side-Chain Transformers

3D Conformational Generative Models for Biological Structures Using Graph Information-Embedded Relative Coordinates

3DMolNet: A Generative Network for Molecular Structures

Swallowing the Bitter Pill: Simplified Scalable Conformer Generation

Data-efficient generation of protein conformational ensembles with backbone-to-side chain transformers

ET-Flow: Equivariant Flow-Matching for Molecular Conformer Generation

Molecular Conformation Generation via Shifting Scores

Learning to evolve structural ensembles of unfolded and disordered proteins using experimental solution data

Physics-informed generative model for drug-like molecule conformers

Direct Molecular Conformation Generation