Drug target prediction method based on multiple similarity network walk

Abstract

The invention relates to a drug target prediction method based on multiple similarity network walk. The method comprises the following steps of S1, obtaining the interaction relationship between drugsand drugs, targets and targets, drugs and diseases, targets and diseases, drugs and side effects from relevant databases, and constructing corresponding adjacency matrix; S2, by calculating Jaccard similarity among nodes in the adjacency matrix, constructing similarity networks corresponding to each adjacency matrix; S3, using a node2vec method to walk and train each similarity network to generate corresponding network feature vectors; S4, splicing the feature vectors of each network obtained in S3 are to obtain a combined feature vector of multiple networks of the drugs and targets; S5, according to an existing drug target relationship, obtaining a positive sample of a drug target pair, generating a negative sample equivalent to the positive sample by randomly combining, and splicing thedrug feature vectors and the target feature vectors obtained in S4 according to the combination of the positive and negative samples, thereby obtaining a final feature vector of the drug target pair;S6, using a random forest algorithm to conduct training, ten-fold cross-check and classification prediction on the positive and negative samples in S5.

Description

technical field [0001] The invention relates to the interdisciplinary field of bioinformatics and computer science, and more specifically, relates to a drug target prediction method based on multiple similarity network walks. Background technique [0002] Predicting the drug-target relationship with methods in the computer field has become a very important step in the process of discovering new drugs and drug repositioning. The potential drug target relationship identified by machine learning can provide guidance for biochemical or clinical experiments, thereby greatly reducing the time and cost of biochemical experiments. [0003] In the traditional field of machine learning, the extraction and selection of features is a very critical part, and the quality of feature representation usually determines the performance of machine learning methods. The selection of features needs to be done effectively by experts in specific fields. Therefore, when conducting interdisciplinary...

AI Technical Summary

Problems solved by technology

[0008] In order to solve the technical defect that the drug target prediction method provided by the prior art ignores the relationship between the drug and the target network and the network topology characteristics, the prediction accuracy is low, and it provides a drug based on multiple similarity network walks. Target prediction method, compared with the traditional drug target prediction method, this method more fully mines the features contained in each similarity network, and provides more sufficient network information for the prediction classifier, thereby improving the accuracy of prediction

Images