Coupled matrix-matrix and coupled tensor-matrix completion methods for predicting drug-target interactions.

Area of Science:

• Computational biology
• Bioinformatics
• Drug discovery and development

Background:

• Accurate prediction of drug-target interactions (DTIs) is essential for efficient drug discovery and repurposing.
• Experimental determination of DTIs is time-consuming and expensive, necessitating advanced computational methods.
• Matrix factorization techniques have shown promise in predicting DTIs.

Purpose of the Study:

• To develop novel and efficient computational methods for predicting drug-target interactions (DTIs).
• To improve the accuracy and speed of DTI prediction by leveraging comprehensive data and advanced algorithms.
• To introduce Coupled Matrix-Matrix Completion (CMMC) and Coupled Tensor-Matrix Completion (CTMC) for enhanced DTI prediction.

Main Methods:

• Proposed a matrix factorization-based method, Coupled Matrix-Matrix Completion (CMMC).
• Extended CMMC to Coupled Tensor-Matrix Completion (CTMC) by incorporating drug-drug and target-target similarity/interaction tensors.
• Utilized comprehensive information from different databases and multiple score types for drug-drug and target-target relationships.

Main Results:

• CTMC demonstrated superior performance compared to existing matrix factorization methods (GRMF, $L_{2,1}$-GRMF, NRLMF, NRLMF$\beta$) on DrugBank and TTD datasets.
• Both CMMC and CTMC achieved higher area under the curve, F1 score, sensitivity, and specificity than baseline methods.
• The proposed methods significantly reduced the computational runtime for DTI prediction.

Conclusions:

• CMMC and CTMC represent effective computational approaches for predicting drug-target interactions.
• These methods offer a more efficient and accurate alternative to traditional experimental approaches for DTI identification.
• The tensor-based extension (CTMC) further enhances prediction accuracy by utilizing richer similarity information.