Title: Drug-target interaction prediction using deep belief network

Authors: Aman Shakya; Basanta Joshi; Uday K. Yadav; Om Prakash Mahato

Addresses: Department of Electronics and Computer Engineering, Institute of Engineering, Tribhuvan University, Pulchowk Campus, Kathmandu, 44600, Nepal ' Department of Electronics and Computer Engineering, Institute of Engineering, Tribhuvan University, Pulchowk Campus, Kathmandu, 44600, Nepal ' Computer Department at Nepal College of Information Technology (NCIT), Pokhara University, Kathmandu, 44600, Nepal ' Electronics and Computer Department at Thapathali Campus, Institute of Engineering, Tribhuvan University, Kathmandu, 44600, Nepal

Abstract: Virtual screening is applied for identifying potential drug-target interactions in drug discovery. There is a need to reduce the search space by identifying ligands that are non-dockable. It is complex to classify dockable and non-lockable ligands as the feature space is too high. Machine learning can be used to efficiently classify drug-target pairs. In this paper, a new framework is proposed to predict the interaction between drugs and targets using deep belief network (DBN). DBN is used to extract high level features from 2D chemical substructures represented in fingerprint format. DBN is trained in a greedy layer-wise unsupervised fashion and the result from this pre-training phase is used to initialise the parameters used for fine tuning. Logistic regression layer is stacked as an output layer. It is shown that this DBN model improves the throughput by two fold with around 90% accuracy for drug and target interaction prediction.

Keywords: drug-target interaction; high throughput screening; virtual screening; RBM; restricted Boltzmann machine; Gibbs sampling; MRF; Markov random field; DBN; deep belief network; back propagation.

DOI: 10.1504/IJBRA.2022.128248

International Journal of Bioinformatics Research and Applications, 2022 Vol.18 No.5, pp.479 - 495

Received: 01 Mar 2022
Accepted: 21 Sep 2022
Published online: 12 Jan 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article