Improve Accuracy of Peptide Identification with Consistency between Peptides

A new method is presented to estimate the accuracy of peptide identification with logistic regression (LR) based on Sequest scores. Each peptide is characterized with the regularized Sequest scores ΔCn* and Xcorr*. The score regularization is formulated as an optimization problem by applying two assumptions: the smoothing consistency between sibling peptides and the fitting consistency between original scores and new scores. An adjacency matrix is built to describe the affinity between peptides, and is used in the score regularization to compute new scores. Then, the new scores are input to the LR model, which is solved with the penalized Newton Raphson method. By applying the method on two datasets with known validity, the results have shown that the proposed method can robustly assign accurate probabilities to peptides and have a very high discrimination power, higher than that of PeptideProphet, to distinguish correct and incorrect peptides.