Prediction Of Drug Targets Based On Machine Learning

When drugs are administrated, they will interact with certain proteins to combat the diseases. The drug target proteins can be classified into on-targets and off-targets, where the off-targets may lead to severe side effects. Therefore, it is necessary to identify the therapeutic targets of drugs so that the desirable therapeutic effects can be guaranteed. In this thesis, several works on predicting drug targets are presented.Firstly, we present the recent progresses on computational methodologies that have been developed to identify drug targets. In particular, we focus on those methodologies based on gene expression data, molecular networks, and pharmacological information due to the rich resources of these types of data.Secondly, we develop an ensemble classifier, PTEC(Predicting Therapeutic targets with Ensemble Classifier), that can efficiently integrate both drug and protein properties described from distinct perspectives, thereby improving prediction accuracy. The results on benchmark datasets demonstrate that our approach outperforms other popular approaches significantly, implying the effectiveness of our proposed approach. Furthermore, the results indicate that the integration of different data sources can not only improve the coverage of predicted targets but also the prediction precision.Lastly, we present a new computational approach to predict targetable protein-protein interactions(PPIs). Based on the properties of amino acids in protein-protein interaction interface, protein structure information, protein-protein interaction and drug targets, a new machine learning approach is developed. Furthermore, with feature extraction and feature selection, some redundant and irrelevant features are removed, thereby improving the prediction accuracy. Results on real data demonstrate that our proposed approach can accurately predict the targetable PPIs.