Neighborhood Regularized Logistic Matrix Factorization for Drug-Target Interaction Prediction

• DOI: 10.1371/journal.pcbi.1004760
• Publication Date: 2016-02-12T18:57:00Z
• Authors: Yong Liu; Min Wu; Chunyan Miao; Peilin Zhao; Xiao-Li Li
• Keywords: Drug target; Benchmark (surveying); Drug repositioning; Non-negative Matrix Factorization; Drug Development

In pharmaceutical sciences, a crucial step of the drug discovery process is identification drug-target interactions. However, only small portion interactions have been experimentally validated, as experimental validation laborious and costly. To improve efficiency, there great need for development accurate computational approaches that can predict potential to direct verification. this paper, we propose novel interaction prediction algorithm, namely neighborhood regularized logistic matrix factorization (NRLMF). Specifically, proposed NRLMF method focuses on modeling probability would interact with target by factorization, where properties drugs targets are represented drug-specific target-specific latent vectors, respectively. Moreover, assigns higher importance levels positive observations (i.e., observed interacting pairs) than negative unknown pairs). Because already verified, they usually more trustworthy. Furthermore, local structure data has also exploited via regularization achieve better accuracy. We conducted extensive experiments over four benchmark datasets, demonstrated its effectiveness compared five state-of-the-art approaches.