An adaptive approach to denoising tandem mass spectra

In our recently developed denoising method, a linear combination of five features is used to adjust the peak intensities in tandem mass spectra. Although the method shows a promise, the coefficients (weights) of the linear combination were fixed and determined empirically. In this paper, we propose an adaptive approach for estimating these weights. The proposed approach: (1) calculates the score for each peak in a data set with the empirically determined weights in, (2) selects the training dataset based on the scores of peaks, (3) applies the LDA (Linear discriminant analysis) to the training dataset and take the solution of LDA as the new weights, (4) calculates the score again with new weights, (5) repeats (2) - (4) until weights have no significant change. After getting the final weights, the proposed approach follows the methods developed in. The proposed approach is applied to two tandem mass spectra datasets: ISB (with low resolution) and TOV-Q (with high resolution) to evaluate its performance. The results show that about 66% of peaks (likely noise peaks) can be removed and that the number of peptides identified by Mascot increases by 14% and 21% for ISB and TOV-Q dataset, respectively, comparing to the previous work.