WeChat Mini Program
Old Version Features

SE(3)-Stochastic Flow Matching for Protein Backbone Generation

ICLR 2024(2024)

Postdoc

Cited 70|Views39
Abstract
The computational design of novel protein structures has the potential toimpact numerous scientific disciplines greatly. Toward this goal, we introduceFoldFlow, a series of novel generative models of increasing modeling powerbased on the flow-matching paradigm over $3\mathrm{D}$ rigid motions -- i.e.the group $\text{SE}(3)$ -- enabling accurate modeling of protein backbones. Wefirst introduce FoldFlow-Base, a simulation-free approach to learningdeterministic continuous-time dynamics and matching invariant targetdistributions on $\text{SE}(3)$. We next accelerate training by incorporatingRiemannian optimal transport to create FoldFlow-OT, leading to the constructionof both more simple and stable flows. Finally, we design FoldFlow-SFM, couplingboth Riemannian OT and simulation-free training to learn stochasticcontinuous-time dynamics over $\text{SE}(3). Our family of FoldFlow, generativemodels offers several key advantages over previous approaches to the generativemodeling of proteins: they are more stable and faster to train thandiffusion-based approaches, and our models enjoy the ability to map anyinvariant source distribution to any invariant target distribution over$\text{SE}(3)$. Empirically, we validate FoldFlow, on protein backbonegeneration of up to $300$ amino acids leading to high-quality designable,diverse, and novel samples.
More
Translated text
Key words
Proteins,Equivariance,Riemannian,Flow Matching,Generative models
PDF
Bibtex
AI Read Science
Must-Reading Tree
Example
Generate MRT to find the research sequence of this paper
Data Disclaimer
The page data are from open Internet sources, cooperative publishers and automatic analysis results through AI technology. We do not make any commitments and guarantees for the validity, accuracy, correctness, reliability, completeness and timeliness of the page data. If you have any questions, please contact us by email: report@aminer.cn
Chat Paper
Summary is being generated by the instructions you defined