Wavelet–Based Denoising of Liquid Chromatography–Mass Spectrometry Data

We present a new method for rejecting noise from LC–MS data sets. The algorithm reveals peptides at low concentrations by minimizing both the chemical and the random noise. The goal is reached through a systematic approach to characterize and remove the background. The data are represented as two-dimensional maps, in order to optimally exploit the complementary dimensions of separation of the peptides offered by the LC-MS technique. The virtual chromatograms, reconstructed from the spectrometric data, have proved to be more suitable to characterize the noise than the raw mass spectra. By means of wavelet analysis, it was possible to access both the chemical and the random noise, at different scales of the decomposition. The novel approach has proved to unveil low-intensity peptides originally masked by the chemical noise and to reduce false positive identification, by filtering noise peaks originally mimicking the peptide morphology. The filtering strategy has been tested in a standard workflow for label-free LC-MS differential analysis. The results showed that the inclusion of a preprocessing step of background subtraction in a common laboratory pipeline can lead to an enhanced inclusion list of peptides selected for fragmentation and, consequently, to better protein identification.