Prediction of drug–target interaction networks from the integration of chemical and genomic spaces

Abstract

Motivation: The identification of interactions between drugs and target proteins is a key area in genomic drug discovery. Therefore, there is a strong incentive to develop new methods capable of detecting these potential drug–target interactions efficiently.Results: In this article, we characterize four classes of drug–target interaction networks in humans involving enzymes, ion channels, G-protein-coupled receptors (GPCRs) and nuclear receptors, and reveal significant correlations between drug structure similarity, target sequence similarity and the drug–target interaction network topology. We then develop new statistical methods to predict unknown drug–target interaction networks from chemical structure and genomic sequence information simultaneously on a large scale. The originality of the proposed method lies in the formalization of the drug–target interaction inference as a supervised learning problem for a bipartite graph, the lack of need for 3D structure information of the target proteins, and in the integration of chemical and genomic spaces into a unified space that we call ‘pharmacological space’. In the results, we demonstrate the usefulness of our proposed method for the prediction of the four classes of drug–target interaction networks. Our comprehensively predicted drug–target interaction networks enable us to suggest many potential drug–target interactions and to increase research productivity toward genomic drug discovery.Availability: Softwares are available upon request.Contact: Yoshihiro.Yamanishi@ensmp.frSupplementary information: Datasets and all prediction results are available at http://web.kuicr.kyoto-u.ac.jp/supp/yoshi/drugtarget/.

Combining Drug and Gene Similarity Measures for Drug-Target Elucidation

“…(i) The kernel regression-based method (KRM) of Yamanishi et al (2008) embeds drugs and targets into a unified Euclidean space termed the ''pharmacological space,'' using a regression model. Predicted interacting drug-gene pairs are those that are closer to each other below a certain threshold in the pharmacological space.…”

“…We note that dopamine receptor D3 (DRD3), having only two known interactions in our data, had exceptionally high occurrence in our novel prediction set (28 times). Dopamine receptor D3 is primarily predicted to be targeted by drugs indicated for Parkinson's disease and schizophrenia (eight out of (Yamanishi et al, 2008) 0.838 0.884 BLM (Bleakley and Yamanishi, 2009) 0.754 0.814 nine existing in DrugBank). It is noteworthy that half of the predicted drugs targeting DRD3 are also indicated for Parkinson's disease or schizophrenia.…”

“…These methods suffer from the complex and noisy nature of molecular networks. Recently, several algorithms have been proposed to predict drug-target associations by combining drug-drug and gene-gene similarity measures (Bleakley and Yamanishi, 2009;Campillos et al, 2008;Keiser et al, 2009;Yamanishi et al, 2008). The key assumption underlying these algorithms is that similar drugs tend to share similar targets (Mitchell, 2001).…”

“…Several authors had previously predicted drug-target interactions by combining chemical drug-drug similarity and sequence-based gene-gene similarity (Bleakley and Yamanishi, 2009;Yamanishi et al, 2008). Keiser et al (2009) compared the chemical structure of drugs to a compendium of ligands, known to modulate the function of protein receptors, providing indirect connections between drugs and targets via these ligands.…”

Integrative methods for analyzing big data in precision medicine

“…The second group of methods uses similarities between diseases (e.g., phenotype similarity [151], or similarity between disease symptoms [152]) to group diseases and to infer a novel drug for repurposing by expanding known associations between the drug and some members of the group to the rest of the group. Other approaches use target-based similarities [153], i.e., protein sequence similarity [154], or 3D structural similarity [155], to infer novel drugs. On the other hand, all three approaches can be classified as similarity-based approaches [153].…”