Fig. 2

Principal component analysis can be used to distinguish antibody repertoires. a. PCA was applied to features of V-J gene use frequency for three healthy donors (HIP1–3). Artificial replicates were generated by subsampling each repertoire to a common depth, repeated 10 times per repertoire. X and Y axes show principal components 1 and 2, and the percent variation explained by each component is shown in parenthesis. b. PCA was applied to gene use frequencies from three healthy donors (HIP1–3) and five HIV-infected donors after influenza vaccination (HIV/Flu1–5). Black circles show results of K-means clustering with k = 2 clusters