Immunological Diversity with Similarity

A diverse immune repertoire is considered a hallmark of good health. However, other things being equal, current methods for measuring repertoire diversity cannot distinguish between a repertoire composed of similar sequences (clonotypes, clones) and a repertoire composed of different ones, even though the latter is intuitively more diverse. Here we describe a framework for incorporating similarity into diversity measures, and measure immunological diversity with similarity on 391 large-scale antibody and T-cell receptor (TCR) repertoires. We find that while repertoires often contain millions of unique sequences, diversity with similarity suggests a landscape defined by at most a few thousand unrelated CDR3 binding targets. Naïve/IgM repertoires have more unique sequences than memory/IgG, but with similarity, memory/IgG repertoires are more diverse. Diversity with similarity is sensitive to vaccination, infection, and aging, and unlike diversity without similarity is robust to sampling error. Finally, with similarity, repertoires from different people overlap significantly, suggesting a definable ceiling for the functional diversity of humanity. Similarity redefines diversity in complex systems.