Compound function prediction method based on neural network and connection graph algorithm

Abstract

The invention provides a compound function prediction method based on a neural network and a connection graph algorithm, or a compound efficacy prediction method. The method comprises the following steps: constructing a compound molecule-coding vector neural network; constructing and training a coding vector-marker gene expression quantity change deep neural network; constructing and training a neural network for mapping marker gene expression quantity or expression variation quantity to whole genome gene expression quantity or gene expression variation quantity; constructing a disease or phenotype up-regulation and down-regulation gene set; and evaluating the correlation of the compounds to diseases or phenotypes. The compound function prediction method provided by the invention is provided on the basis of a neural network and a connection graph algorithm, can realize high-flux prediction of expression change of a compound on genes in cells in the early stage of drug research and development, and can predict functions and side effects of the compound according to the expression change of the genes; therefore, the drug research and development process can be greatly accelerated.

Description

technical field [0001] The invention relates to medicine informatics and artificial intelligence, in particular to a medicine function evaluation method based on neural network and gene enrichment algorithm. Background technique [0002] According to statistics, it takes 10-14 years and more than 200 million U.S. dollars for each new drug to go on the market, from testing to marketing. How to speed up the discovery and testing of new drugs has always been the key and difficulty in accelerating the drug R&D stage. In recent years, the development of biochemistry, physiopathology and other disciplines has provided new means of drug screening, and some drug screening models at the molecular and cellular level have emerged, and with the development of more advanced detection technology, automation technology and computer technology, In the late 1990s, a high throughput screening technology (High throughput screening, HTS) was developed. HTS mainly relies on automated operating ...

AI Technical Summary

Problems solved by technology

However, the existing connectivity map (CMAP) technology is only applicable to small molecule compounds with 1309 known data points, and cannot make predictions for other molecules, especially virtual molecules

Images