Predicting Drug-Target Interactions with Neighbor Interaction Information and Discriminative Low-rank Representation

Inferring drug-target interaction (DTI) candidates for new drugs or targets without any interaction information is a critical challenge for modern drug design and discovery. A few approaches are applied to solve this problem. Results from these applications indicate that the existing DTI inference methods necessitate further improvement. The sparse, low-rank, and nonnegative properties of a known DTI matrix prompted us to design a novel DTI identification model ( i.e, PreNNDS ) by integrating known neighbor interaction profiles, nonnegative matrix factorization, discriminative low-rank representation, and sparse representation classification into a unified framework. Experimental results on the four types of DTI networks show that PreNNDS can efficiently identify potential DTIs for new drugs or targets. Further analysis of the predicted results demonstrates that the predicted DTIs deserve further biomedical experimental validation. PreNNDS can be applied to identify multi-target drugs and multi-drug resistance proteins, as well as to provide clues for microRNA-disease and gene-disease association prediction.