Fig. 5

Optimization of MME window for X!Tandem on the hybrid ion trap/maXis dataset described above, with fitness defined at the number of correctly identified spectra from doubly charged precursor divided by the total number of tandem mass spectra. The surface was interpolated and visualized outside Taverna using gnuplot with the dgrid3d and countour base functions, although similar graphics could in the future possibly also be created in an Rshell inside the Taverna workflow. The clearly visible ridge between 16 and 21 Da positive MME corresponds to the pyroglutamate/ammonia loss resonance adding 437 peptide-spectrum matches (4 % of all PSMs) with 1 % FDR and MMEs 15-20 Da, nearly all by assigning actual NH₃-loss b-ions as regular b-ions with 17 Da MME. As comparison, there are only 9 PSMs in the MME window between 10 and 15 Da. Similar ridges are seen in at least four of the six datasets (supplemental information), although the global optimum is not always found along this ridge. It should be noted that the standard error in the actual mass measurement is below 2 ppm in this dataset, but that this number has very little relevance for the optimum MME window for X!Tandem in a search of this dataset with only one variable modification and in a small sequence database