Verifying the interaction between drug and target is a key step to find new drug. Utilizing computational methods to predict drug-target interaction (DTI) is the most efficient and skillful way, but a lot of challenges also exist in the identification of DTI. Currently, a lot of in silico studies on DTIs were developed based on binary classification, which oversimplifies the complex relationships between drugs and targets and makes the prediction results largely dependent on the threshold of positive and negative samples. This paper proposed an integrative network fusion and machine learning to predict drug-target binding affinity, called INM-PBA, which applied multiple network information in the prediction of specific binding affinity who represents the binding strength between drug and target. As a result, our model obtained excellent results and outperformed the state-of-the-art methods. 

Website: http://deeplearner.ahu.edu.cn/web/inmpba.htm

drug-target datasets and supplementary files: supplementary.