TROLLOPE: A novel sequence-based stacked approach for the accelerated discovery of linear T-cell epitopes of hepatitis C virus

Abstract

Hepatitis C virus (HCV) infection is a concerning health issue that causes chronic liver diseases. Despite many successful therapeutic outcomes, no effective HCV vaccines are currently available. Focusing on T cell activity, the primary effector for HCV clearance, T cell epitopes of HCV (TCE-HCV) are considered promising elements to accelerate HCV vaccine efficacy. Thus, accurate and rapid identification of TCE-HCVs is recommended to obtain more efficient therapy for chronic HCV infection. In this study, a novel sequence-based stacked approach, termed TROLLOPE, is proposed to accurately identify TCE-HCVs from sequence information. Specifically, we employed 12 different sequence-based feature descriptors from heterogeneous perspectives, such as physicochemical properties, composition-transition-distribution information and composition information. These descriptors were used in cooperation with 12 popular machine learning (ML) algorithms to create 144 base-classifiers. To maximize the utility of these base-classifiers, we used a feature selection strategy to determine a collection of potential base-classifiers and integrated them to develop the meta-classifier. Comprehensive experiments based on both cross-validation and independent tests demonstrated the superior predictive performance of TROLLOPE compared with conventional ML classifiers, with cross-validation and independent test accuracies of 0.745 and 0.747, respectively. Finally, a user-friendly online web server of TROLLOPE (http://pmlabqsar.pythonanywhere.com/TROLLOPE) has been developed to serve research efforts in the large-scale identification of potential TCE-HCVs for follow-up experimental verification.

Fig 1. System flowchart of the proposed TROLLOPE.

The development and performance assessment of TROLLOPE involves five main steps: dataset preparation, feature representation, stacked model development, performance evaluation, and online web server deployment.

Fig 2. Performance evaluations of top-30 base-classifiers.

(A-B) Cross-validation AUC and ACC of top-30 base-classifiers. (C-D) Independent AUC and ACC of top-30 base-classifiers.

Fig 3

Performance comparison of TROLLOPE and top-five base-classifiers on the training (A–B) and independent (C–D) datasets.

Fig 4. t-distributed stochastic neighbor embedding (t-SNE) distribution of positive and negative samples on the training dataset, where TCE-HCV and non-TCE-HCV are represented with red and blue dots, respectively.

TROLLOPE (A) and top-five base-classifiers (B-F).

Fig 5

t-SNE plots of our new feature OPF (A) and top-three feature descriptors (B-D) (i.e. DDE, DPC, and TPC) on the training dataset.

Fig 6. Feature importance analysis for TROLLOPE prediction.

(A) Scatter plot of top-15 informative probabilistic features. (B) The average absolute SHAP values of top-15 informative probabilistic features.

Fig 7

Feature importance analysis for XGB-AAI (A-B) and XGB-PCP (C-D) predictions. (A, C) Scatter plot of top 20 informative features. (B, D) The average absolute SHAP values of top 20 informative features.