Methods comparison for sharing prediction.

(A) Distribution of the sharing number of CDR3 amino-acid sequences of IgG repertoires from 10 individuals. The prediction from the model (green line) comes from sequences simulated using the immuneSIM software [21]. The simulation dramatically diverges from the real amount of shared sequences indicating either a bad recombination model or an overestimation of selection. (B) Distribution of the sharing number of CDR3 amino-acid sequences of TCRβ from 666 patients. Model prediction (green line) was obtained using the default model of vampire [23] trained on the same data set. vampire slightly overestimates the sharing distribution.

(TIFF)