DTI-HeNE: a novel method for drug-target interaction prediction based on heterogeneous network embedding

Yang Yue,Shan He

doi:10.1186/s12859-021-04327-w

Yang Yue, Shan He

Open Access

https://doi.org/10.1186/s12859-021-04327-w

Copy DOI

Abstract

BackgroundPrediction of the drug-target interaction (DTI) is a critical step in the drug repurposing process, which can effectively reduce the following workload for experimental verification of potential drugs’ properties. In recent studies, many machine-learning-based methods have been proposed to discover unknown interactions between drugs and protein targets. A recent trend is to use graph-based machine learning, e.g., graph embedding to extract features from drug-target networks and then predict new drug-target interactions. However, most of the graph embedding methods are not specifically designed for DTI predictions; thus, it is difficult for these methods to fully utilize the heterogeneous information of drugs and targets (e.g., the respective vertex features of drugs and targets and path-based interactive features between drugs and targets).ResultsWe propose a DTI prediction method DTI-HeNE (DTI based on Heterogeneous Network Embedding), which is specifically designed to cope with the bipartite DTI relations for generating high-quality embeddings of drug-target pairs. This method splits a heterogeneous DTI network into a bipartite DTI network, multiple drug homogeneous networks and target homogeneous networks, and extracts features from these sub-networks separately to better utilize the characteristics of bipartite DTI relations as well as the auxiliary similarity information related to drugs and targets. The features extracted from each sub-network are integrated using pathway information between these sub-networks to acquire new features, i.e., embedding vectors of drug-target pairs. Finally, these features are fed into a random forest (RF) model to predict novel DTIs.ConclusionsOur experimental results show that, the proposed DTI network embedding method can learn higher-quality features of heterogeneous drug-target interaction networks for novel DTIs discovery.

Highlights

Prediction of the drug-target interaction (DTI) is a critical step in the drug repurposing process, which can effectively reduce the following workload for experimental verification of potential drugs’ properties
Problem formulation In our study, the DTIs prediction can be formulated as a transductive-learning binary link-prediction task based on a heterogeneous network, which is divided into a bipartite DTI network as well as drug and target homogeneous networks
In order to tackle the problem that some recent embedding-based methods cannot add the pathway information about drug-target interactions into embeddings of drug-target pairs, we provide a method, which draws on the path-based information, to acquire new embeddings of every drug-target pair (i.e., the reconstruction of DTI relations included in the whole dataset)

Summary

Introduction

Prediction of the drug-target interaction (DTI) is a critical step in the drug repurposing process, which can effectively reduce the following workload for experimental verification of potential drugs’ properties. A recent trend is to use graph-based machine learning, e.g., graph embedding to extract features from drug-target networks and predict new drug-target interactions. DTIs prediction based on computational techniques plays an important role in drug repurposing because it requires lower cost and less time, compared with biochemical experimental methods [2,3,4]. There are two varieties of traditional computational methods: the ligand-based method [6] and the structure-based or docking-based method [7], which can provide relatively accurate DTI predictions. The former one has the limitation on predictive performance when few binding ligands are provided for a certain target, while the latter will not be feasible when the three-dimensional (3D) structure of the target is not available [2]

Methods

Results

Conclusion