Affinity2Vec: drug-target binding affinity prediction through representation learning, graph mining, and machine learning

Abstract

Drug-target interaction (DTI) prediction plays a crucial role in drug repositioning and virtual drug screening. Most DTI prediction methods cast the problem as a binary classification task to predict if interactions exist or as a regression task to predict continuous values that indicate a drug's ability to bind to a specific target. The regression-based methods provide insight beyond the binary relationship. However, most of these methods require the three-dimensional (3D) structural information of targets which are still not generally available to the targets. Despite this bottleneck, only a few methods address the drug-target binding affinity (DTBA) problem from a non-structure-based approach to avoid the 3D structure limitations. Here we propose Affinity2Vec, as a novel regression-based method that formulates the entire task as a graph-based problem. To develop this method, we constructed a weighted heterogeneous graph that integrates data from several sources, including drug-drug similarity, target-target similarity, and drug-target binding affinities. Affinity2Vec further combines several computational techniques from feature representation learning, graph mining, and machine learning to generate or extract features, build the model, and predict the binding affinity between the drug and the target with no 3D structural data. We conducted extensive experiments to evaluate and demonstrate the robustness and efficiency of the proposed method on benchmark datasets used in state-of-the-art non-structured-based drug-target binding affinity studies. Affinity2Vec showed superior and competitive results compared to the state-of-the-art methods based on several evaluation metrics, including mean squared error, rm2, concordance index, and area under the precision-recall curve.

Drugs–Protein affinity‐score prediction using deep convolutional neural network

“…Two different datasets have been used to evaluate the model. Earlier, the benchmark datasets used for the evaluation of binding affinity prediction were the KIBA dataset and kinase Davis dataset (Öztürk et al, 2018;Öztürk et al, 2019;Shim et al, 2021;Thafar et al, 2022;Wen et al, 2017;Zeng et al, 2021).…”

“…As new drugs are discovered and added to the field of drug discovery, there is more interest in repurposing existing drugs and finding new ways for approved drugs to work together (Oprea & Mestres, 2012). Predicting DTI was treated as a binary classification problem (Bleakley & Yamanishi, 2009; Cao et al, 2012; Cao et al, 2014; Cobanoglu et al, 2013; Gönen, 2012; Öztürk et al, 2016; Öztürk et al, 2018; Thafar et al, 2022). This meant that the binding affinity values were not considered, which essential information about how proteins and ligands interact is.…”

Machine learning empowered drug discovery

“…Exscientia-Sanofi collaborative AI-assisted research identified a novel molecule for fibrosis and is in clinical trials [22]. Affinity2Vec, a KronRLS (Kronecker-regularized least squares) based algorithm developed to assess the drug-target binding affinity (DTBA) [23].…”

OncoRTT: Predicting novel oncology-related therapeutic targets using BERT embeddings and omics features

“…There are also other models focused on cancer-related drug repurposing that predict drug response in cancer cell lines ( Liu et al, 2020 ) and novel oncology drug-target interactions (DTIs) ( Huang et al, 2016 ; Dezső and Ceccarelli, 2020 ). In addition, other groups have proposed more generic DTIs prediction methods ( Thafar et al, 2020b ; Thafar et al, 2020 ; Alshahrani et al, 2021 ; Alshahrani et al, 2022 ; Thafar et al, 2022 ) with high prediction performance that provides similar topic-specific information ( Thafar et al, 2019 ). All these avenues could lead to artificial intelligence (AI) tools that support clinicians and pinpoint potential new drugs.…”

“…Here, we contribute to this line of research by developing the target protein prediction method, OncoRTT, that better exploits efficient features of the known targets using more advanced approaches and integrating features from several resources to improve target protein prediction in a topic-specific manner (more importantly, specific cancer types). Thus, our method, OncoRTT, is the first attempt to use DL-based models whose primary goal is to systematically predict potential cancer-specific therapeutic targets ( Thafar, 2022 ). The main contributions of this work can be summarized as follow.…”