Figure 3
From: MUMAL: Multivariate analysis in shotgun proteomics using machine learning techniques
Example showing how the inclusion of other parameters can improve sensitivity in a decoy DB approach. For a FDR = 0, just three identifications can be retrieved when applying Xcorr thresholds. The addition of ΔC n allows threshold combinations resulting in five identifications (rectangles and ellipses). (a) Textual representation. (b) Graphical representation (ΔC n vs. Xcorr), where crosses represent decoy PSMs and circles denote normal PSMs. We added here the green curve in (b) to show that a non-linear decision boundary is expected to further enhance sensitivity in such analysis.