Figure 5

Reliable ranking of (A) CDR3 and (B) VDJ sequences depends on deep sequence coverage of samples. Plots show the ratio of reliably ranked clones-expressed as the median number of clones having a rank coefficient of variation (CV) lower than 0.05 to the number of reliably detected clones determined in Figure 3 as a function of the proportion of accumulated reads. A LOESS curve was fitted to the calculated ratios (displayed as points) in order to extrapolate their overall distribution with a 0.95 confidence interval (gray-shaded area). As species richness increases, the maximum value of the ratio decreases. Across diversity scenarios (1M/9M) and clonal definitions (CDR3/VDJ) the respective maximum is reached at 25% of accumulated reads. Simulations were performed analogously to those shown in Figure 2 using a sequential sampling scheme (bootstrapping). The CV was calculated based on ranks of CDR3/VDJ clones across triplicates. Absolute numbers of reliably ranked clones were: CDR3: 730 (1M), 4160 (9M); VDJ: 3708 (1M), 5169 (9M).