PeBLes: Prediction of B-cell epitope using molecular layers

TL;DR

This paper introduces PeBLes, a novel molecular layer-based method for predicting B-cell epitopes by identifying anchor residues, optimizing sampling strategies, and integrating multiple features to achieve high accuracy in epitope prediction.

Contribution

The study presents a new approach using the Layers algorithm to identify anchor residues and optimize surface sampling for improved B-cell epitope prediction.

Findings

Majority of epitopes are composed of anchor residues.

Optimized sampling reduces molecular surface to 75 residues while retaining 50% of the epitope.

Achieved 89% accuracy using combined molecular features.

Abstract

Characterization of B-cell protein epitope and developing critical parameters for its identification is one of the long standing interests. Using Layers algorithm, we introduced the concept of anchor residues to identify epitope. We have shown that majority of the epitope is composed of anchor residues and have significant bias in epitope for these residues. We optimized the search space reduction for epitope identification. We used Layers to non-randomly sample the antigen surface reducing the molecular surface to an average of 75 residues while preserving 50% of the epitope in the sample surface. To facilitate the comparison of favorite methods of researchers we compared the popular techniques used to identify epitope with their sampling performance and evaluation. We proposed an optimum Sr of 16 {\AA} to sample the antigen molecules to reduce the search space, in which epitope is identified using buried surface area method. We used the combinations of molecular surface sampling, anchor residue intensity in surface, secondary structure and sequence information to predict epitope at an accuracy of 89%. A web application is made available at http://www.csb.iitkgp.ernet.in/applications/b_cell_epitope_pred/main.