2022

Equivariant Diffusion for Molecule Generation in 3D

Hoogeboom, Emiel, Satorras, Victor Garcia, Vignac, Clément et al.

Understand

This work introduces a diffusion model for molecule generation in 3D that is equivariant to Euclidean transformations.

Our E(3) Equivariant Diffusion Model (EDM) learns to denoise a diffusion process with an equivariant network that jointly operates on both continuous (atom coordinates) and categorical features (atom types).
In addition, we provide a probabilistic analysis which admits likelihood computation of molecules using our model.
Experimentally, the proposed method significantly outperforms previous 3D molecular generative methods regarding the quality of generated samples and efficiency at training time.

Built on

Generative modeling by estimating gradients of the data distribution
Original
Song, Y. and Ermon, S · 1907
Earlier work this paper cites.
Linear representations of finite groups , volume 42
Serre, J.-P · 1977
Earlier work this paper cites.
Quantum chemistry structures and properties of 134 kilo molecules
Ramakrishnan, R., Dral, P. O., Rupp, M., and Von Lilienfeld, O. A · 2014
Earlier work this paper cites.
Deep unsupervised learning using nonequilibrium thermodynamics
Sohl-Dickstein, J., Weiss, E. A., Maheswaranathan, N., and Ganguli, S · 2015
Earlier work this paper cites.
Neural message passing for quantum chemistry
Gilmer, J., Schoenholz, S. S., Riley, P. F., Vinyals, O., and Dahl, G. E · 2017
Earlier work this paper cites.
Molgan: An implicit generative model for small molecular graphs
De Cao, N. and Kipf, T · 2018
Earlier work this paper cites.
Generating equilibrium molecules with deep neural networks
Original
Gebauer, N. W., Gastegger, M., and Schütt, K. T · 2018
Earlier work this paper cites.
Constrained graph variational autoencoders for molecule design
Original
Liu, Q., Allamanis, M., Brockschmidt, M., and Gaunt, A. L · 2018
Earlier work this paper cites.
Graphvae: Towards generation of small graphs using variational autoencoders
Simonovsky, M. and Komodakis, N · 2018
Earlier work this paper cites.
Tensor field networks: Rotation- and translation-equivariant neural networks for 3d point clouds
Original
Thomas, N., Smidt, T., Kearnes, S. M., Yang, L., Li, L., Kohlhoff, K., and Riley, P · 2018
Earlier work this paper cites.
Graph convolutional policy network for goal-directed molecular graph generation
Original
You, J., Liu, B., Ying, R., Pande, V., and Leskovec, J · 2018
Earlier work this paper cites.
Cormorant: Covariant molecular neural networks
Anderson, B., Hy, T. S., and Kondor, R · 2019
Earlier work this paper cites.
Gfn2-xtb—an accurate and broadly parametrized self-consistent tight-binding quantum chemical method with multipole electrostatics and density-dependent dispersion contributions
Bannwarth, C., Ehlert, S., and Grimme, S · 2019
Earlier work this paper cites.
A two-step graph convolutional decoder for molecule generation
Original
Bresson, X. and Laurent, T · 2019
Earlier work this paper cites.
Symmetry-adapted generation of 3d point sets for the targeted discovery of molecules
Original
Gebauer, N. W., Gastegger, M., and Schütt, K. T · 2019
Earlier work this paper cites.

Similar

Generating valid euclidean distance matrices
Original
Hoffmann, M. and Noé, F · 2019
Cited alongside, same era.
Efficient graph generation with graph recurrent attention networks
Original
Liao, R., Li, Y., Song, Y., Wang, S., Nash, C., Hamilton, W. L., Duvenaud, D., Urtasun, R., and Zemel, R. S · 2019
Cited alongside, same era.
Boltzmann generators: Sampling equilibrium states of many-body systems with deep learning
Noé, F., Olsson, S., Köhler, J., and Wu, H · 2019
Cited alongside, same era.
A generative model for molecular distance geometry
Original
Simm, G. N. and Hernández-Lobato, J. M · 2019
Cited alongside, same era.
Geom: Energy-annotated molecular conformations for property prediction and molecular generation
Original
Axelrod, S. and Gomez-Bombarelli, R · 2020
Cited alongside, same era.
Generalizing convolutional neural networks for equivariance to lie groups on arbitrary continuous data
Finzi, M., Stanton, S., Izmailov, P., and Wilson, A. G · 2020
Cited alongside, same era.
Se(3)-transformers: 3d roto-translation equivariant attention networks
Fuchs, F., Worrall, D. E., Fischer, V., and Welling, M · 2020
Cited alongside, same era.
Denoising diffusion probabilistic models
Original
Ho, J., Jain, A., and Abbeel, P · 2020
Cited alongside, same era.
Directional message passing for molecular graphs
Klicpera, J., Groß, J., and Günnemann, S · 2020
Cited alongside, same era.
Equivariant flows: Exact likelihood generative learning for symmetric densities
Köhler, J., Klein, L., and Noé, F · 2020
Cited alongside, same era.
Conditional set generation with transformers
Kosiorek, A. R., Kim, H., and Rezende, D. J · 2020
Cited alongside, same era.
Learning a continuous representation of 3d molecular structures with deep generative models
Original
Ragoza, M., Masuda, T., and Koes, D. R · 2020
Cited alongside, same era.

Then

Inverse design of 3d molecular structures with conditional generative neural networks
Original
Gebauer, N. W., Gastegger, M., Hessmann, S. S., Müller, K.-R., and Schütt, K. T · 2021
Later among the works it cites.
Argmax flows and multinomial diffusion: Learning categorical distributions
Hoogeboom, E., Nielsen, D., Jaini, P., Forré, P., and Welling, M · 2021
Later among the works it cites.
Highly accurate protein structure prediction with alphafold
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al · 2021
Later among the works it cites.
Variational diffusion models
Original
Kingma, D. P., Salimans, T., Poole, B., and Ho, J · 2021
Later among the works it cites.
Gg-gan: A geometric graph generative adversarial network, 2021
Krawczuk, I., Abranches, P., Loukas, A., and Cevher, V · 2021
Later among the works it cites.
An autoregressive flow model for 3d molecular geometry generation from scratch
Luo, Y. and Ji, S · 2021
Later among the works it cites.
A graph vae and graph transformer approach to generating molecular graphs
Original
Mitton, J., Senn, H. M., Wynne, K., and Murray-Smith, R · 2021
Later among the works it cites.
Improved denoising diffusion probabilistic models
Original
Nichol, A. and Dhariwal, P · 2021
Later among the works it cites.
Learning gradient fields for molecular conformation generation
Shi, C., Luo, S., Xu, M., and Tang, J · 2021
Later among the works it cites.
Symmetry-aware actor-critic for 3d molecular design
Simm, G. N. C., Pinsler, R., Csányi, G., and Hernández-Lobato, J. M · 2021
Later among the works it cites.
Top-n: Equivariant set and graph generation without exchangeability
Original
Vignac, C. and Frossard, P · 2021
Later among the works it cites.
Energy-inspired molecular conformation optimization
Guan, J., Qian, W. W., qiang liu, Ma, W.-Y., Ma, J., and Peng, J · 2022
Closest in time.
Progressive distillation for fast sampling of diffusion models
Original
Salimans, T. and Ho, J · 2022
Closest in time.
Geodiff: A geometric diffusion model for molecular conformation generation
Xu, M., Yu, L., Song, Y., Shi, C., Ermon, S., and Tang, J · 2022
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…