Skip to main content
. Author manuscript; available in PMC: 2018 May 1.
Published in final edited form as: Res Comput Mol Biol. 2017 Apr 12;10229:99–116. doi: 10.1007/978-3-319-56970-3_7

Fig. 2. Averaged TDC (a-TDC). A-D.

Fig. 2

Comparing the FDR controlling procedure of a-TDC with 1 (TDC), 3, 10, and 100 competing decoys. E-F a-TDC with 10 competing decoys. A Plotted are the log of the ratios of the mean (empirical FDR, middle curves) as well as the 0.05 and 0.95 quantiles (upper and lower curves) of the FDP in the target discovery lists of each of the four procedures, to the nominal FDR level. Scores are calibrated. B The 0.05 and 0.95 quantiles of the number of target discoveries. Scores are calibrated. C Same as panel A, except the simulations were done using the raw (uncalibrated) score. D Shown are the logarithm of the median (middle curves), 0.05 and 0.95 quantiles of the number of true target discoveries reported by a-TDC (with 3, 10, and 100 decoys) over the corresponding number reported by TDC at the same FDR threshold. Scores are uncalibrated. E The log of the median of the ratio of the number of true a-TDC to TDC discoveries show that the power advantage of a-TDC over TDC diminishes with the increase in the number of calibrating decoys. F Coincidentally, a-TDC becomes less conservative: the middle set of curves show that the log of the empirical FDR (mean of FDP) over the nominal level increases toward 0 for small FDR levels. (The 0.05 and 0.95 quantiles are also provided.) A-F All quantiles are taken with respect to 10K simulations, each with 10K spectra, 50% native spectra.