TransPhos: A Deep-Learning Model for General Phosphorylation Site Prediction Based on Transformer-Encoder Architecture

Abstract

Protein phosphorylation is one of the most critical post-translational modifications of proteins in eukaryotes, which is essential for a variety of biological processes. Plenty of attempts have been made to improve the performance of computational predictors for phosphorylation site prediction. However, most of them are based on extra domain knowledge or feature selection. In this article, we present a novel deep learning-based predictor, named TransPhos, which is constructed using a transformer encoder and densely connected convolutional neural network blocks, for predicting phosphorylation sites. Data experiments are conducted on the datasets of PPA (version 3.0) and Phospho. ELM. The experimental results show that our TransPhos performs better than several deep learning models, including Convolutional Neural Networks (CNN), Long-term and short-term memory networks (LSTM), Recurrent neural networks (RNN) and Fully connected neural networks (FCNN), and some state-of-the-art deep learning-based prediction tools, including GPS2.1, NetPhos, PPRED, Musite, PhosphoSVM, SKIPHOS, and DeepPhos. Our model achieves a good performance on the training datasets of Serine (S), Threonine (T), and Tyrosine (Y), with AUC values of 0.8579, 0.8335, and 0.6953 using 10-fold cross-validation tests, respectively, and demonstrates that the presented TransPhos tool considerably outperforms competing predictors in general protein phosphorylation site prediction.

PETrans: De Novo Drug Design with Protein-Specific Encoding Based on Transfer Learning

“…With the rapid development of machine learning techniques, quantitative structure activity relationships (QSAR) have become an indispensable virtual screening filter to efficiently and reliably evaluate various physicochemical and pharmacological properties. However, the traditional approach tends to search molecules with desired properties from existing chemical libraries [ 2 , 47 ].…”

“…Traditional methods such as high-throughput screening are inefficient because the number of resources required is not balanced by the small number of hit compounds. Conventionally, the identification of promising lead structures is achieved by experimental high-throughput screening (HTS), but this is time-consuming and expensive [ 1 , 2 , 3 ]. A typical drug discovery cycle takes approximately 14 years [ 4 ] and costs approximately 800 million dollars [ 5 ].…”

Integrating Multiple Single-Cell RNA Sequencing Datasets Using Adversarial Autoencoders

“…The rapid development of high-throughput single-cell RNA sequencing (scRNA-seq) technologies has facilitated the study of the transcriptomic characterization of cell heterogeneity and dynamics [ 1 , 2 , 3 , 4 ]. In recent years, researchers have collected a large amount of single-cell gene expression data from different experiments at different times and on different sequencing platforms [ 5 , 6 ].…”

CAT-CPI: Combining CNN and transformer to learn compound image features for predicting compound-protein interactions

“…MADE ( Pang et al, 2022 ) constructs two different encoders to learn the graph information and sequence information of the drug respectively, and then uses a feature fusion atttention-based method which integating the drug multiple dimensions features. TransPhos ( Wang et al, 2022 ) proposes a two-stage deep learning approach and constructs three different structures of encoders for feature learning based on the attention mechanism. SDNN-PPI ( Li et al, 2022 ) constructs three different ways of encoding protein sequences, and then uses a self-attention mechanism to further learn semantic relationships in the sequences for Protein-Protein Interaction (PPI).…”