Assigning probabilities to Mascot peptide identification using logistic regression.

We propose a method to assign probabilities to Mascot peptide identification by using logistic regression. Three key scores, Mascot ions score (MIS), identity threshold, and homology threshold, are integrated into the logistic regression model. Two features in the model are constructed as the differences between MIS and the two thresholds, respectively. Newton-Raphson method is then adopted to solve the model and the weight vector is estimated by maximizing the likelihood of training data. By applying the method to two datasets with known validity, the results demonstrate that the proposed method can assign accurate probabilities to Mascot peptide identifications and have a high discrimination power to separate correct and incorrect peptide identifications.