Predicting Drug–Target Interactions Using Probabilistic Matrix Factorization

Abstract

Quantitative analysis of known drug–target interactions emerged in recent years as a useful approach for drug repurposing and assessing side effects. In the present study, we present a method that uses probabilistic matrix factorization (PMF) for this purpose, which is particularly useful for analyzing large interaction networks. DrugBank drugs clustered based on PMF latent variables show phenotypic similarity even in the absence of 3D shape similarity. Benchmarking computations show that the method outperforms those recently introduced provided that the input data set of known interactions is sufficiently large—which is the case for enzymes and ion channels, but not for G-protein coupled receptors (GPCRs) and nuclear receptors. Runs performed on DrugBank after hiding 70% of known interactions show that, on average, 88 of the top 100 predictions hit the hidden interactions. De novo predictions permit us to identify new potential interactions. Drug–target pairs implicated in neurobiological disorders are overrepresented among de novo predictions.

Small Random Forest Models for Effective Chemogenomic Active Learning

“…To date, several pro-and retrospective studies report successful applications of AL strategies throughout different fields of research [27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42][43][44][45][46]. In the area of drug discovery, active learning has been shown to efficiently derive high-performance prediction models based on small subsets of input data [32,34,39,46].…”

“…In the area of drug discovery, active learning has been shown to efficiently derive high-performance prediction models based on small subsets of input data [32,34,39,46]. Furthermore, actively trained models not only reached significantly higher hit rates compared to experimental standards which frequently remain below 1 % in cases of unbiased chemical libraries [34,39,40,[47][48][49], but such models also contributed to successful identification of novel bioactive compounds [33,36,42] and cancer rescue mutants of p53 [31]. Overall, AL approaches bear the potential to improve drug discovery processes by increasing hit rates, reducing the amount of time-and cost-intensive experimentation, and accelerate hit-to-lead processes through integration into a feedback-driven experimentation workflow [33,36,42,43,50].…”

Predicting drug-target interaction based on sequence and structure information

“…The chemical structure similarity between compounds is calculated by using SIMCOMP, which gives a score based on the size of common substructures with graph alignment (Hattori et al 2003). The chemical structure similarity has been widely applied in drug-target interaction prediction (Cobanoglu et al 2013).…”

“…The Gaussian interaction profile kernel and weighted nearest neighbour are integrated for drug-target interactions prediction (Van Laarhoven 2013). In addition, the Bayesian matrix factorization and binary classification (Gönen 2012) and probabilistic matrix factorization (Cobanoglu et al 2013) are proposed to detect drug-target interactions.…”

Predicting drug–target interaction using positive-unlabeled learning

“…The Gaussian interaction profile kernel and weighted nearest neighbour were integrated for drug-target interaction prediction [23]. The Bayesian matrix factorization and binary classification [24] and probabilistic matrix factorization [25] were proposed to detect drug-target interactions. The common limitation of these supervised learning approaches is to treat unknown drug-target interactions as negative samples, which may affect predictive accuracy.…”