Predicting MHC class I binder: existing approaches and a novel recurrent neural network solution

• Published in: Briefings in bioinformatics Vol. 22; no. 6
• Main Authors: Jiang, Limin, Yu, Hui, Li, Jiawei, Tang, Jijun, Guo, Yan, Guo, Fei
• Format: Journal Article
• Language: English
• Published: England Oxford University Press 05.11.2021
• Subjects: Amino Acid Sequence; Binding Sites; Carrier Proteins - chemistry; Carrier Proteins - metabolism; Computational Biology - methods; Databases, Factual; Deep Learning; Epitopes - chemistry; Epitopes - immunology; Epitopes - metabolism; Histocompatibility Antigens Class I - chemistry; Histocompatibility Antigens Class I - immunology; Histocompatibility Antigens Class I - metabolism; Machine Learning; Neural Networks, Computer; Protein Binding; Reproducibility of Results; Review; ROC Curve; Software; Web Browser; MHC class I; Position-specific scoring matrix; variable recurrent neural network; long short-term memory
• ISSN: 1467-5463; 1477-4054; 1477-4054
• DOI: 10.1093/bib/bbab216

Major histocompatibility complex (MHC) possesses important research value in the treatment of complex human diseases. A plethora of computational tools has been developed to predict MHC class I binders. Here, we comprehensively reviewed 27 up-to-date MHC I binding prediction tools developed over the last decade, thoroughly evaluating feature representation methods, prediction algorithms and model training strategies on a benchmark dataset from Immune Epitope Database. A common limitation was identified during the review that all existing tools can only handle a fixed peptide sequence length. To overcome this limitation, we developed a bilateral and variable long short-term memory (BVLSTM)-based approach, named BVLSTM-MHC. It is the first variable-length MHC class I binding predictor. In comparison to the 10 mainstream prediction tools on an independent validation dataset, BVLSTM-MHC achieved the best performance in six out of eight evaluated metrics. A web server based on the BVLSTM-MHC model was developed to enable accurate and efficient MHC class I binder prediction in human, mouse, macaque and chimpanzee.