Extended Data Fig. 2: Pseudovirus neutralization of affinity-matured variants.
From: Efficient evolution of human antibodies from general protein language models
(a) Neutralization curves for wildtype antibodies (gray) and variants obtained by our language-model-guided affinity maturation campaigns. Also see Supplementary Tables 5, 8, and 9 for corresponding IC50 values. Points indicate the mean; error bars indicate the standard deviation; n = 4 independent assays. (b) Fold-improvement in kon has low correlation with fold-change in IC50 (Spearman r = 0.12), while fold-improvement in koff has high correlation with fold-change in IC50 (Spearman r = 0.79); compare to Fig. 3c. Correlations involve n = 15 antibody variants. We define a higher kon and a lower koff as improved, so we divide the mutant value by the wildtype value to calculate fold-improvement in kon and vice-versa to calculate fold-improvement in koff.
