Proteins accessible to immune surveillance show significant T-cell epitope depletion: Implications for vaccine design

Vaxign-ML: supervised machine learning reverse vaccinology model for improved prediction of bacterial protective antigens

MOTIVATION

Reverse vaccinology (RV) is a milestone in rational vaccine design, and machine learning (ML) has been applied to enhance the accuracy of RV prediction. However, ML-based RV still faces challenges in prediction accuracy and program accessibility.

RESULTS

This study presents Vaxign-ML, a supervised ML classification to predict bacterial protective antigens (BPAgs). To identify the best ML method with optimized conditions, five ML methods were tested with biological and physiochemical features extracted from well-defined training data. Nested 5-fold cross-validation and leave-one-pathogen-out validation were used to ensure unbiased performance assessment and the capability to predict vaccine candidates against a new emerging pathogen. The best performing model (eXtreme Gradient Boosting) was compared to three publicly available programs (Vaxign, VaxiJen, and Antigenic), one SVM-based method, and one epitope-based method using a high-quality benchmark dataset. Vaxign-ML showed superior performance in predicting BPAgs. Vaxign-ML is hosted in a publicly accessible web server and a standalone version is also available.

AVAILABILITY AND IMPLEMENTATION

Vaxign-ML website at http://www.violinet.org/vaxign/vaxign-ml, Docker standalone Vaxign-ML available at https://hub.docker.com/r/e4ong1031/vaxign-ml and source code is available at https://github.com/VIOLINet/Vaxign-ML-docker.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Epitope distribution in ordered and disordered protein regions - part A. T-cell epitope frequency, affinity and hydropathy

Highly disordered protein regions are prevalently hydrophilic, extremely sensitive to proteolysis in vitro, and are expected to be under-represented as T-cell epitopes. The aim of this research was to find out whether disorder and hydropathy prediction methods could help in understanding epitope processing and presentation. According to the pan-specific T-cell epitope predictors NetMHCpan and NetMHCIIpan and 9 publicly available disorder predictors, frequency of epitopes presented by human leukocyte antigens (HLA) class-I or -II was found to be more than 2.5 times higher in ordered than in disordered protein regions (depending on the disorder predictor). Both HLA class-I and HLA class-II binding epitopes are prevalently hydrophilic in disordered and prevalently hydrophobic in ordered protein regions, whereas epitopes recognized by HLA class-II alleles are more hydrophobic than those recognized by HLA class-I. As regards both classes of HLA molecules, high-affinity binding epitopes display more hydrophobicity than low affinity-binding epitopes (in both ordered and disordered regions). Epitopes belonging to disordered protein regions were not predicted to have poor affinity to HLA class-II molecules, as expected from disorder intrinsic proteolytic instability. The relation of epitope hydrophobicity and order/disorder location was also valid if alleles were grouped according to the HLA class-I and HLA class-II supertypes, except for the class-I supertype A3 in which the main part of recognized epitopes was prevalently hydrophilic. Regarding specific supertypes, the affinity of epitopes belonging to ordered regions varies only slightly (depending on the disorder predictor) compared to the affinity of epitopes in corresponding disordered regions. The distribution of epitopes in ordered and disordered protein regions has revealed that the curves of order-epitope distribution were convex-like while the curves of disorder-epitope distribution were concave-like. The percentage of prevalently hydrophobic epitopes increases with the enhancement of epitope promiscuity level and moving from disordered to ordered regions. These data suggests that reverse vaccinology, oriented towards promiscuous and high-affinity epitopes, is also oriented towards prevalently hydrophobic, ordered regions. The analysis of predicted and experimentally evaluated epitopes of cancer-testis antigen MAGE-A3 has confirmed that the majority of T-cell epitopes, particularly those that are promiscuous or naturally processed, was located in ordered and disorder/order boundary protein regions overlapping hydrophobic regions.

Computer aided selection of candidate vaccine antigens

Immunoinformatics is an emergent branch of informatics science that long ago pullulated from the tree of knowledge that is bioinformatics. It is a discipline which applies informatic techniques to problems of the immune system. To a great extent, immunoinformatics is typified by epitope prediction methods. It has found disappointingly limited use in the design and discovery of new vaccines, which is an area where proper computational support is generally lacking. Most extant vaccines are not based around isolated epitopes but rather correspond to chemically-treated or attenuated whole pathogens or correspond to individual proteins extract from whole pathogens or correspond to complex carbohydrate. In this chapter we attempt to review what progress there has been in an as-yet-underexplored area of immunoinformatics: the computational discovery of whole protein antigens. The effective development of antigen prediction methods would significantly reduce the laboratory resource required to identify pathogenic proteins as candidate subunit vaccines. We begin our review by placing antigen prediction firmly into context, exploring the role of reverse vaccinology in the design and discovery of vaccines. We also highlight several competing yet ultimately complementary methodological approaches: sub-cellular location prediction, identifying antigens using sequence similarity, and the use of sophisticated statistical approaches for predicting the probability of antigen characteristics. We end by exploring how a systems immunomics approach to the prediction of immunogenicity would prove helpful in the prediction of antigens.

Understanding infectious agents from an in silico perspective

Knowledge of infectious diseases now emerging from genomic, proteomic, epidemiological and clinical data can provide insights into the mechanisms of immune function, disease pathogenesis and epidemiology. Here, we describe how considerable advances in computational methods of data mining, mathematical modeling in epidemiology and simulation have been used to enhance our understanding of infectious agents and discuss their impact on the discovery of new therapeutics and controlling their spread.