Fig. 2: Fitness landscape of S proteins of Omicron lineages as of late 2022.

a Epidemic dynamics of S haplotypes in the UK. Omicron sequences were classified into 254 groups harboring unique sets of substations in the S protein, referred to as S haplotypes. S haplotypes are ordered according to the time of epidemic peak. b Effect size of each substitution in the S protein on relative effective reproduction number (R_e) estimated by a hierarchal Bayesian model. The posterior mean value is shown. A group of highly co-occurred substitutions (e.g., L452R and F486V) was treated as substitution clusters. The red and blue dots indicate the substitutions with significant positive and negative effects, respectively. The representative substitutions are annotated. c Relative R_e value for a viral group represented by each S haplotype, assuming a fixed generation time of 2.1 day. The posterior mean value is shown. The R_e of the major S haplotype in BA.2 is set at 1. The substitution profile at the five convergent sites is shown on the left. d Prediction of the relative R_e of S haplotypes in the USA using the model trained on UK data. The predicted R_e and R_e estimated by a simple multiple logistic model based on USA’s data were compared. The dot size indicates the number of sequences of each haplotype. The dotted line denotes a line with a slope of 1 and an intercept of 0. e Adjusted R² value for the prediction of the R_e of S haplotypes in each country. The bar color indicates the total number of sequences included in the dataset investigated. f Comparison between relative R_e and the total effect of substitutions at the convergent sites on R_e. Dot indicates a viral group represented by an S haplotype. The dots are colored according to the major classification of the PANGO lineage. g Change in viral fitness during BA.5 diversification. The lineages indicated with an asterisk, which includes BQ.1.1, are magnified in the right panel.