Predicting Drug Target Interactions Based on GBDT
The research of the drug-target interactions (DTIs) is of great significance for drug development. Traditional chemical experiments are expensive and time-consuming. In recent years, many computational approaches based on different principles have been proposed gradually. Most of them use the information of drug-drug similarity and target-target similarity and made some progress. But the result is far from satisfactory. In this paper, we proposed machine learning method based on GBDT to predict DTIs with the IDs of both drug and protein, the descriptor of them, known DTIs and double negative samples. After gradient boosting and supervised training, GBDT construct decision trees for drug-target networks and generate precise model to predict new DTIs. Experimental results shows that Gradient Boosting Decision Tree (GBDT) reaches or outperforms other state-of-the-art methods.
KeywordsDrug-target interaction prediction DTIs Drug discovery Machine learning GBDT DrugBank
- 5.DrugBank. https://www.drugbank.ca/
- 6.Van, L.T., Nabuurs, S.B., Marchiori, E.: Gaussian Interaction Profile Kernels for Predicting Drug-Target Interaction. Oxford University Press, Oxford (2011)Google Scholar
- 9.Zheng, X., Ding, H., Mamitsuka, H., et al.: Collaborative matrix factorization with multiple similarities for predicting drug-target interactions. In: ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1025–1033. ACM (2013)Google Scholar
- 15.Ye, J., Chow, J.H., Chen, J., et al.: Stochastic gradient boosted distributed decision trees. In: ACM Conference on Information and Knowledge Management, pp. 2061–2064 (2009)Google Scholar
- 16.Chen, T., He, T., Benesty, M.: Xgboost: extreme gradient boosting. R Package Version 0.4-2, 1–4 (2015)Google Scholar