Predicting structures of substances using diffusion models
The substance discovery system employs a diffusion neural network to predict substance structures, addressing the inefficiencies of existing methods by generating diverse and stable structures efficiently, thus reducing validation costs and time.
WO2025072762A1 Publication Date: 2025-04-03GOOGLE LLC
4 Cites 0 Cited by
Patent Information
- Application Number
- PCT/US2024/048980
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-09-28
- Filing Date
- 2024-09-27
- Publication Date
- 2025-04-03
AI Technical Summary
Technical Problem
Existing methods for discovering novel substances are time-consuming and computationally expensive, often requiring hundreds or thousands of human expert hours for validation, and they struggle to generate diverse and stable substance structures.
Method used
A substance discovery system utilizing a diffusion neural network to predict the structure of substances through a reverse diffusion process, enabling the efficient generation of novel, stable, and diverse substance structures.
Benefits of technology
The system efficiently discovers novel and stable substances by generating diverse and high-fidelity substance structures, reducing the time and computational costs associated with laboratory experimentation and validation.
✦ Generated by Eureka AI based on patent content.
Abstract
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for predicting structures of substances. One of the methods includes: generating an initial structure of a substance, wherein the initial structure of the substance defines a respective initial location for each of a plurality of components of the substance; generating a final structure of the substance that defines a respective final location for each of the plurality of components of the substance, wherein updating the initial structure of the substance comprises, at each of multiple updating iterations: processing one or more diffusion inputs for the updating iteration that each comprise a current structure of the substance as of the updating iteration using a diffusion neural network to generate one or more diffusion outputs for the updating iteration; and updating the current structure of the substance using the one or more diffusion outputs for the updating iteration.
Need to check novelty before this filing date? Find Prior Art