Computation of Antigenicity Predicts SARS-CoV-2 Vaccine Breakthrough Variants

AbstractIt has been reported that multiple SARS-CoV-2 variants of concerns (VOCs) including B.1.1.7 (Alpha), B.1.351 (Beta), P.1 (Gamma), and B.1.617.2 (Delta) can reduce neutralisation by antibodies, resulting in vaccine breakthrough infections. Virus-antiserum neutralisation assays are typically performed to monitor potential vaccine breakthrough strains. However, such experimental-based methods are slow and cannot instantly validate whether newly emerging variants can break through current vaccines or therapeutic antibodies. To address this, we sought to establish a computational model to predict the antigenicity of SARS-CoV-2 variants by sequence alone and in real time. In this study, we firstly identified the relationship between the antigenic difference transformed from the amino acid sequence and the antigenic distance from the neutralisation titres. Based on this correlation, we obtained a computational model for the receptor binding domain (RBD) of the spike protein to predict the fold decrease in virus-antiserum neutralisation titres with high accuracy (~0.79). Our predicted results were comparable with experimental neutralisation titres of variants, including B.1.1.7 (Alpha), B.1.351 (Beta), B.1.617.2 (Delta), B.1.429 (Epsilon), P.1 (Gamma), B.1.526 (Iota), B.1.617.1 (Kappa), and C.37 (Lambda), as well as SARS-CoV. Here, we firstly predicted the fold of decrease of B.1.1.529 (Omicron) as 17.4-fold less susceptible to neutralisation. We visualised all 1521 SARS-CoV-2 lineages to indicate variants including B.1.621 (Mu), B.1.630, B.1.633, B.1.649, and C.1.2, which can induce vaccine breakthrough infections in addition to reported VOCs B.1.351 (Beta), P.1 (Gamma), B.1.617.2 (Delta), and B.1.1.529 (Omicron). Our study offers a quick approach to predict the antigenicity of SARS-CoV-2 variants as soon as they emerge. Furthermore, this approach can facilitate future vaccine updates to cover all major variants. An online version can be accessed at http://jdlab.online.