Drug recommendation system based on genetic algorithm integrated deep learning network, computer equipment and storage medium

Abstract

The invention discloses a drug recommendation system based on a genetic algorithm integrated deep learning network, and belongs to the technical fields of drug relocation, deep learning and integrated learning model prediction. The system comprises an integrated deep learning network based on a genetic algorithm, and the integrated deep learning network based on the genetic algorithm comprises a feature extraction network, genetic algorithm optimization weights and a sequence processing model composed of a full-connection network. According to the integrated deep learning network based on the genetic algorithm, a drug molecule SMILES sequence and a target protein FASTA sequence are input into the feature extraction network for sufficient feature extraction so as to obtain feature vectors; the genetic algorithm is used for distributing weights for the multiple feature vectors; the feature vectors are multiplied by the weights of the feature vectors and then spliced together to form spliced vectors; the spliced vectors and a true value of the drug-target binding affinity are input into the full-connection network for regression analysis; predicted values of the drug-target binding affinity are output and sorted; and finally a drug molecule recommendation list which integrates multiple feature representation modes and aims at a specific target is obtained.

Description

technical field [0001] The invention relates to the technical field of genetic algorithm and deep learning model prediction, in particular to a drug recommendation system, computer equipment and storage medium based on genetic algorithm integrated deep learning network. Background technique [0002] The prediction of drug-target binding affinity is the key to drug repositioning. The drug molecular sequence, target protein amino acid sequence and drug-target binding affinity value are input into the deep learning network for training, and the trained network is used to predict unknown drugs - The target binding affinity value has become one of the methods with the highest prediction accuracy. [0003] However, the prediction results of each deep learning model are different, and the coincidence rate of the prediction results of different deep learning models is low. When selecting drug molecules for biological experiment verification, there are still too many choices and a hi...

AI Technical Summary

Problems solved by technology

[0003] However, the prediction results of each deep learning model are different, and the coincidence rate of the prediction results of different deep learning models is low. When selecting drug molecules for biological experiment verification, there are still too many choices and a high failure rate.

Images