Prediction of Drug-Target Interactions by Ensemble Learning Method From Protein Sequence and Drug Fingerprint

Predicting the target-drug interactions (DITs) is of great important for screening new drug candidate and understanding biological processes. However, identifying the drug-target interactions through traditional experiments is still costly, laborious and complicated. Thus, there is a great need for developing reliable computational methods to effectively predict DTIs. In this study, we report a novel computational method combining local optimal oriented pattern (LOOP), Position Specific Scoring Matrix (PSSM) and Rotation Forest (RF) for predicting DTI. Specifically, the target protein sequence is firstly transformed as the PSSM, in which the evolutionary information of protein is retained. Then, the LOOP is used to extract the feature vectors from PSSM, and the sub-structure information of drug molecule is represented as fingerprint features. Finally, RF classifier is adopted to infer the potential drug-target interactions. When the experiment is carried out on four benchmark datasets including enzyme, ion channel, Gprotein-coupled receptors (GPCRs), and nuclear receptor, we achieved the high average prediction accuracies of 89.09%, 87.53%, 82.05%, and 73.33% respectively. For further evaluating the proposed method, we compare the prediction performance of the proposed method with the state-of-the-art support vector machine (SVM) and K-Nearest Neighbor (KNN). The comprehensive experimental results illustrate that the proposed method is reliable and efficiency for predicting DTIs. It is anticipated that the proposed method can become a useful tool for predicting a large-scale potential DTIs.

Abstract licence: CC BY-NC-ND

1. 13.

   Semi-supervised drug-protein interaction prediction from heterogeneous biological spaces · 2010

2. 34.

   Neighborhood Regularized Logistic Matrix Factorization for Drug-Target Interaction Prediction · 2016

38 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.