Patent · US Active

Neural network for predicting drug property

US10923214B2 · kind B2 · utility

5Cited by
0References
20Claims
0Family size

Assignee

Inventors

Key dates

Filing dateSep 7, 2018
Grant dateFeb 16, 2021
Priority date
Expiry dateMay 9, 2039

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG06N3/02
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

A computer-implemented method for predicting molecule properties is disclosed. According to some embodiments, the method may include receiving an input file of a compound. The method may also include implementing a neural network to determine molecular configurations of the compound based on the input file and a plurality of molecular descriptors associated with the compound. The method may also generating, using the neural network, one or more three-dimensional (3D) models of the compound based on the determined molecular configurations of the compound. The method may also include determining, using the neural network, energy scores of the one or more 3D models when the compound is docked into a protein. The method may further include determining a property of the docked compound based on the energy scores.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.