Figure 7
Determination of peptide microarray performance using ROC analysis. The data displayed in Fig. 6, panels A and B, were used to determine performance characteristics of the C. trachomatis and C. pneumoniae microarray peptides for antibody detection against C. trachomatis or C. pneumoniae in human sera determined by receiver operating characteristic (ROC) curves and area under curve (AUC). Red lines indicate binormal ROC curves derived by maximum likelihood estimation, black lines indicate 95% confidence intervals. (A) ROC analysis of continuous RSI results of C. trachomatis-specific antibody detection with OmpA, PmpD, and CT529 peptides in comparison to positive/negative scoring by anti-C. trachomatis ELISA. (B) ROC analysis of C. pneumoniae-specific antibody detection with IncA and PmpD peptides in comparison to positive/negative scoring by anti-C. pneumoniae ELISA. (C) ROC analysis of ordinal C. trachomatis ELISA results in comparison to positive/negative scoring by anti-C. trachomatis RSI data with a negative RSI cut-off of <0.2 RSI. (D) ROC analysis of ordinal C. pneumoniae ELISA results in comparison to positive/negative scoring by anti-C. pneumoniae RSI data with <0.2 RSI cut-off. Data for this panel are plotted directly since a binormal ROC curve could not be constructed due to an implied exact-fit that is vertical at constant FPF = 0.091.
