Unsupervised clustering analysis of B-cell profiles segregates lupus patients into distinct groups. Flow cytometry data from 25 healthy controls (HC) and 137 systemic lupus erythematosus (SLE) patients were clustered independently by B-cell phenotypic profiles using Matlab (MathWorks, Natick MA, USA). Clustering was based on Euclidean distance and complete linkage using a reduced feature set to avoid correlated cell subsets based on the gating strategy. Subset frequencies (in rows) were logit-transformed and each cell subset was standardized to its mean and standard deviation of all 162 samples (in columns) prior to clustering. This approach segregated lupus patients into three distinct clusters, and representative lupus patients from each cluster were shown. Preliminary analysis indicates that SLE-I cluster is enriched for patients with high Systemic Lupus Erythematosus Disease Activity Index and high serum interferon alpha activity, in contrast to SLE-II cluster whose B-cell profile resembles that of healthy controls (manuscript in preparation). Note that the B-cell profiles among the healthy controls are relatively heterogeneous. Subset frequencies are the percentages of total B cells, unless indicated otherwise. CD19+ frequencies are percentages of lymphocytes. DN, double negative; N + T, IgD+CD27− fraction that contains both naïve and transitional cells; SM, switched memory; T, CD24++CD38++ transitional B cells; UM, unswitched memory.