Fig 1.
Distribution of B. pseudomallei-challenged and negative sera screened with the BurkPx assay.
Forty-seven B. pseudomallei negative sera collected from routine cross-colony surveys of non-human primates at TNPRC were screened with this assay (black bar, first column) to serve as B. pseudomallei cross-reactivity controls. The distribution of serum samples (n = 115) collected from 80 rhesus macaques involved in the inhalational LD50 determination experiment are shown over time. Days pre- or post-challenge and total number of sera samples obtained at each time point are shown in parentheses. The different challenge strains of B. pseudomallei associated with each sample are indicated by the colored stacked bars.
Table 1.
B. pseudomallei proteins, carbohydrates, and assay controls comprising the BurkPx multiplex serological assay.
Fig 2.
Antigen reactivity for B. pseudomallei-challenged and B. pseudomallei-negative NHP sera.
Heat maps are shown displaying antigen reactivity for IgM (top panel) and IgG (bottom panel). Burkholderia antigens are shown in rows and columns represent individual serum samples. Samples included negative sera (n = 126 sample, n = 126 animals), B. pseudomallei aerosol challenged sera (n = 115 samples, n = 68 animals) and serum samples were further stratified by prescreening or time after B. pseudomallei challenge. Antigens were arranged in order by signal to noise ratio and were ordered differently for each isotype. Sera were run in duplicate and the average median fluorescence intensity values (MFI) are shown. IgM ranges were: 0 (blue), 5,000 (white), 10,000 (yellow), and >10,000 (red) as indicated. IgG ranges were: 0 (blue), 10,000 (white), 20,000 (yellow), and >20,000 (red) as indicated.
Table 2.
AUC values illustrating the fit of a single antigen model for IgM and IgG BurkPx antigens.
Fig 3.
Principal Component Analysis (PCA) and group significance plots of antigen reactive IgM and IgG.
Principle Component Analyses (PCA) were performed using Euclidean distance matrices, which were based upon average median fluorescence intensity of all antigens (MFI) independently for IgM (A) and IgG (B) antibodies. The first two principal components were used to plot individual sera in both cases. Permutational multivariate analysis of variance (PERMANOVA) illustrates pairwise distances within and between groups for IgM (C) and IgG (D). “ns” = not significant, * = q <0.005.
Fig 4.
Receiver Operator Characteristic (ROC) plots and box and whisker plots of the IgGM Lasso multiple antigen models.
-scores (p-hat) were generated from median fluorescent intensity values of all antigens for IgM and IgG. (A and C) Independent test validation and (B and D) cross-validation performance LASSO models are shown.