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

Abstract

Predicting the interactions between drugs and targets plays an important role in the process of new drug discovery, drug repurposing (also known as drug repositioning). There is a need to develop novel and efficient prediction approaches in order to avoid the costly and laborious process of determining drug–target interactions (DTIs) based on experiments alone. These computational prediction approaches should be capable of identifying the potential DTIs in a timely manner. Matrix factorization methods have been proven to be the most reliable group of methods. Here, we first propose a matrix factorization-based method termed ‘Coupled Matrix–Matrix Completion’ (CMMC). Next, in order to utilize more comprehensive information provided in different databases and incorporate multiple types of scores for drug–drug similarities and target–target relationship, we then extend CMMC to ‘Coupled Tensor–Matrix Completion’ (CTMC) by considering drug–drug and target–target similarity/interaction tensors. Results: Evaluation on two benchmark datasets, DrugBank and TTD, shows that CTMC outperforms the matrix-factorization-based methods: GRMF, $L_{2,1}$-GRMF, NRLMF and NRLMF$\beta $. Based on the evaluation, CMMC and CTMC outperform the above three methods in term of area under the curve, F1 score, sensitivity and specificity in a considerably shorter run time.

Data Fusion by Matrix Completion for Exposome Target Interaction Prediction

“…When working with problems including data points in n -dimensional space, □ n , Mahalanobis distance , as the inverse of covariance matrix of the data, learns a metric induced by the data points itself. As proposed in Bagherian M, et al 20 , CMMC method development is heavily based on algebraic property of symmetric matrices which can be thought of elements of General Linear Group (GL) and Special Linear Group (SL). These linear groups belong to a class of groups, called reductive groups 25 .…”

“…The method benefits from strong results in group action and representation theory and guarantees a unique solution for the optimization problem. A different implementation of CMMC using a binary matrix was shown to outperform alternative approaches in the related task of predicting drug target interactions 20 . In this study compared with previously published methods, CMMC achieved better performance on the benchmark datasets.…”

Drug repositioning in drug discovery of T2DM and repositioning potential of antidiabetic agents

“…The DeepDTIs model allows researchers to search for all possible drug-target interactions in the network space, without the limitations of previous target classification. Maryam et al [60] proposed a model named the Coupled Tensor-Matrix Completion (CTMC), which combined matrix factorization-based method with drug-drug and target-target tensors to repurpose drug molecules, and showed a better output in comparison with other matrix-factorization-based methods in performance evaluations. Bai et al [61] , [62] developed a software named MolAICal that could study the interaction between targets and ligands by deep learning models and classical algorithms [63] .…”

An Ensemble Learning-Based Method for Inferring Drug-Target Interactions Combining Protein Sequences and Drug Fingerprints

“…Then, the scores of inner products of factors are trained to predict DTIs. Maryam et al [ 15 ] developed an effective model named the Coupled Tensor-Matrix Completion (CTMC) to repurpose drug molecules by constructing drug-drug and target-target tensors. Pliakos and Vens [ 16 ] proposed to address DTI prediction as a multioutput prediction task by learning ensembles of multioutput biclustering trees (eBICT) on reconstructed networks.…”