Figure 1From: A cross-validation scheme for machine learning algorithms in shotgun proteomicsElimination of confounding variables in PSM scoring. Approximately 30,000 top-scoring PSMs were obtained from a set of C. elegans spectra using a ± 3 Da Sequest search. A target database of C. elegans protein sequences and a separate decoy database of the reversed sequences were used. Each PSM obtained a target and a decoy score, indicated in the 2D-histograms on the x and y axis, respectively. The black line represents the x = y diagonal. (A) shows the score distribution of PSMs when using Sequest's XCorr. (B) shows the same PSMs when scored with Percolator score. The PSM count in each 2D-bin is indicated by color coding.Back to article page