| In recent years,drug discovery and drug repositioning is a challenging task,drug discovery has been widely concerned by researchers.The research content of this thesis is drug-target interaction prediction based on machine learning.We extract the molecular descriptors from the drug structure and target protein amino acid sequence information.Because the feature matrix contains redundant information,therefore,Laplacian score feature selection and three different feature dimensionality reduction methods are used to preprocess the feature matrix,then we put the preprocessed features into three machine learning classifiers for modeling.The AUC value of the random forest classifier is 0.895.After a case study,it is also confirmed that our algorithm has applicability and accuracy. |
