Time-varying filtering of time-of-flight mass spectra for proteomics

Proteomics is the post-genomic science that aims to identify and characterize the entire protein complement of a cell or organism. Mass spectrometry followed by database search is the core technology used in high-throughput protein identification. A major challenge in Proteomics is the random and chemical noise that corrupts mass spectra. This noise can alter the measured mass-to-charge ratio (m/z) and bias peak intensities, thus leading to errors in peptide ion peak detection. Thus, noise reduction by filtering is necessary prior to data analysis.As mass spectrometer response varies with m/z, an invariant filter would not be optimal; therefore, a time-varying filter was developed to denoise mass spectra. This involved conversion of spectra from the m/z domain to the time domain equivalent, modeling spectral peak shape and Fourier-transforming the model to determine its frequency content. The process was completed by filter design and implementation.The application of this filter to experimental spectra resulted in improved signal-to-noise ratio (by factor of 45), enabled detection of low abundance peptide ions, and increased the accuracy of peptide mass assignment