Fig. 2

Performance comparison in terms of PSM identifications using 11 cell line datasets. Four PSM rescoring methods were applied—Percolator, TIDD (iterative SVM learning with tool-independent feature set), TIDD with tool-dependent features (iterative SVM learning with TIDD features augmented with tool-dependent scores), and the iterative SVM using X!Tandem-Percolator features (iterative SVM learning with the feature set used by Percolator on X!Tandem data, while ‘deltascore’ is missed because MSFragger does not provide this score). a the number of identified PSMs based on TD when the three tools were applied; b–d the percent increase in PSM identification numbers compared to TD results