Separation of healthy, SCT, and SCD RBC samples based on lectin specificity of microarrays. (A) RBC samples (ghost cells) were applied to a lectin array (GlycoTechnica) and trends interpreted from fluorescence data. Lectins are colored based in binding specificity (for full specificities see supplemental Table 2). (B) Data were analyzed using the centroids distance method and separated based on disease state (component contributions are listed in supplemental Figure 1). Background shading represents the 95% confidence interval (CI). Receiver operating characteristic values are as follows: Healthy vs all = 0.8046; SCT vs all = 0.6153; SCD vs all = 0.9206. (C) Data normalized using a quantile method and with t test statistics performed showed a number of significant changes (P ≤ .05) in lectin binding between all sample types (95% CI bars shown; for full t test results, see supplemental Table 3). Lectin specificities and disease state preferences for each lectin are indicated.