Introducing a novel computational framework for B-cell epitope prediction based on immune-induced selection signatures

Project Summary The accurate prediction of B-cell epitopes is essential for disease analysis, diagnostic tests, and vaccine design, yet it lags behind T-cell epitope prediction in precision. This is due to the complex nature of B-cell epitopes and the undersampling of mapped antibody-antigen structural data. To address this, we propose calculating the intensity of immune selection as an indicator of immunodominance because immune response-driven selection leaves detectable genetic signatures around common regions targeted by antibodies. Our preliminary data suggests that surface-adapted immune selection statistics recover common epitope sites in Sars-Cov-2 Spike, Influenza HA1, and malaria antigens. We have set out two primary aims to enhance B-cell epitope prediction. Aim 1: we plan to build a novel and comprehensive antigen database from 68 human pathogens, which maps 3D immune selection profiles onto the surfaces of antigens from common pathogens. Antigens will be selected from IEDB and their underlying population- level variation will be extracted from public genomic resources. Population genetics scores, such as Tajima's D and BetaScan, will be calculated on antigen surfaces. The resulting database will be deployed online for easy public access. Aim 2: we will develop a B-cell epitope predictor with two innovations. First, the training output comes from the normalized selection statistics of antigen surface instead of relying on antigen-antibody structures. Second, structural features for training inputs will be encoded through the Holographic-CNN model. The predictor's efficacy will then be compared against the state-of-the-art models using a distinct test set of experimentally resolved antigen-antibody structures. Upon completion, our predictor is expected to substantially elevate the predictive power of B-cell epitopes. Our antigen selection database will provide unparalleled new information for various research purposes of antigen evolution. These tools will be instrumental for reverse vaccinology, especially the design of epitope-based vaccines and the evaluation of the potential effectiveness of immunological interventions.