Supplementary Figure 14: FDR comparison of Prosit and MS2PIP on Bekker-Jensen Trypsin and Bekker-Jense Chymotrypsin.
From: Prosit: proteome-wide prediction of peptide tandem mass spectra by deep learning
Number of estimated true positive (#targets - #decoys at respective false discovery rate (FDR) cutoff) peptide spectrum matches using percolator at different peptide level FDR cutoffs when using the Andromeda (blue), full score set based on intensity predictions (orange) (see Supplementary Table 5 for feature set description). Dashed line indicates the number of true positive identifications when using the Andromeda feature set at 1% peptide level FDR. Top figures show the analysis for trypsin and bottom figures for chymotrypsin. On the left MS2PIP predictions were used and on the right Prosit predictions.
