Clonal tree frequency analysis reveals affinity determinants and enables antibody lead optimization

Understanding the sequence determinants of antibody affinity within a clonal family is a central challenge in therapeutic antibody discovery. Here, we applied amino acid frequency analysis to a large phylogenetic clonal tree of anti-ST2 monoclonal antibodies, where V(D)J recombination is constant and only somatic hypermutations vary. We identified M33I and L100I in the heavy chain as key affinity-governing mutations, corroborated by AlphaFold2 structural modeling and molecular dynamics free‑energy decomposition. The I33/I100 genotype predicted significantly improved functional neutralization of IL-33-induced IL-6 release and was associated with a higher transcriptomic affinity score (Hiscore). Prospective mutagenesis of three low-affinity parent antibodies achieved up to 10-fold affinity improvement, establishing a workflow integrating sequence statistics, structural modeling, transcriptomics and MD simulation for rational lead optimization.