Fig. 3.

RDI accounts for repertoire size heterogeneity. TRB sequences from a single donor in the Rubelt et al. dataset were randomly assigned to one of two unevenly-sized groups. The smaller group contained 1000, 2500, 5000, 10,000, or 50,000 total sequences, and all remaining sequences were assigned to the second group. V gene frequencies from the two repertoires were compared using the RDI method. The distribution of RDI values across 1000 replicates (black histogram) was compared with simulated data (grey curves) with controlled levels of variance (average fold change of gene segments = 1, 1.2, or 1.5; indicated numbers)