Fig. 3
From: Statistical inference of a convergent antibody repertoire response to influenza vaccine
Identifying putative TIV-responding V genes. a First eigenfunction in the B cell data for each patient and each V gene class. The proportion of the total variance explained by the first eigenfunction is listed in the legend after each respective class of V gene. b The top five scoring IGHV genes from the FPCA based test to identify TIV-responding V genes; in the B cell data. The points show the observed data, and the solid lines show the best fitting gene expression function over time. V genes in legend are ordered by p value, with lowest on top. P values are based on a permutation test, see Wu and Wu [28] for details. c, d Same as (a, b) but from the PBMC data. Colors corresponding to IGHV genes in (b, d) are comparable within a patient. Eigenfunction plots (a, c) were generated using the “eigens” output from the FPCA-based test. IGHV gene expression functions (b, d) were plotted using the “fda” package for the R programming language and using a smoothing parameter, “lambda” = 0.66