DTI-BERT: Identifying Drug-Target Interactions in Cellular Networking Based on BERT and Deep Learning Method

Abstract

Drug–target interactions (DTIs) are regarded as an essential part of genomic drug discovery, and computational prediction of DTIs can accelerate to find the lead drug for the target, which can make up for the lack of time-consuming and expensive wet-lab techniques. Currently, many computational methods predict DTIs based on sequential composition or physicochemical properties of drug and target, but further efforts are needed to improve them. In this article, we proposed a new sequence-based method for accurately identifying DTIs. For target protein, we explore using pre-trained Bidirectional Encoder Representations from Transformers (BERT) to extract sequence features, which can provide unique and valuable pattern information. For drug molecules, Discrete Wavelet Transform (DWT) is employed to generate information from drug molecular fingerprints. Then we concatenate the feature vectors of the DTIs, and input them into a feature extraction module consisting of a batch-norm layer, rectified linear activation layer and linear layer, called BRL block and a Convolutional Neural Networks module to extract DTIs features further. Subsequently, a BRL block is used as the prediction engine. After optimizing the model based on contrastive loss and cross-entropy loss, it gave prediction accuracies of the target families of G Protein-coupled receptors, ion channels, enzymes, and nuclear receptors up to 90.1, 94.7, 94.9, and 89%, which indicated that the proposed method can outperform the existing predictors. To make it as convenient as possible for researchers, the web server for the new predictor is freely accessible at: https://bioinfo.jcu.edu.cn/dtibert or http://121.36.221.79/dtibert/. The proposed method may also be a potential option for other DITs.

A Robust Drug–Target Interaction Prediction Framework with Capsule Network and Transfer Learning

“…However, accurately predicting the complete 3D structure of a protein solely from its sequence using BERT-based models remains a complex and challenging problem [ 17 , 38 , 39 , 40 , 41 ]. While BERT-based models are not specifically designed to provide high-resolution 3D structural predictions similar to AlphaFold [ 42 ], they showed promise in capturing local structural features such as secondary structure (alpha-helices and beta-sheets), solvent accessibility, and other properties that can be inferred from sequence patterns [ 4 , 43 ]. In this study, we employ the ProtBert, a large protein language model, to extract contextual features from protein sequences utilizing transfer learning ( Figure 1 B).…”

“…Structure-based methods and ligand-based methods are the two main existing computational approaches for identifying drug–target interactions [ 4 ]. In structure-based strategies such as molecular docking, the three-dimensional (3D) structures of proteins and chemical compounds are utilized to explore potential binding poses at the atom level and identify binding affinities [ 5 ].…”

Machine Learning Advances in Predicting Peptide/Protein‐Protein Interactions Based on Sequence Information for Lead Peptides Discovery

“…PPI databases can be migrated into PepPI predictions to address the lack of training samples due to inefficiencies and false positives, also known as the sparse distribution of training samples. [39,40] Currently, there are several available databases that provide PPIs and PepPIs and most of them are made up of sequence information, reference literature information, interactions, taxonomy, annotation, etc. This section introduces the most commonly used public databases of PPIs and PepPIs, and presents a brief list in Table 1.…”

Stacking-ac4C: an ensemble model using mixed features for identifying n4-acetylcytidine in mRNA

“…Once the benchmark dataset has been prepared for the study, the next important step is formulating the samples and extracting the best feature set for constructing a robust and superior computational model. In recent years, various feature encoding strategies have been used to form biological sequence fragments, such as PseKNC (17), One-hot (21,22), physicochemical features, and word2vec (23)(24)(25). This study selected some of the most common feature encoding approaches, including six physicochemical feature encoding strategies and the frequency of occurrence of k-nearest neighbor nucleic acids, to describe RNA fragments.…”