PNNL

Abstract

Spectral peak detection algorithms are difficult to fully automate because these algorithms most often rely on somewhat arbitary, case specific rules. Manay generic methods detect peaks where the absolute or differenced signal intensity exceeds some threshold. This threshold is typically set arbitrarily above the noise level or manually by the user. For applications such as matrix-assisted laser desorption/ionization (MALDI) mass spectrometry, the peak intensitives, baseline, and noise level tend to vary widely over the mass range of interest. Consequently, intensity threshold-based methods ten to require user interaction where the analyst sets different thresholds for different spectral regions and then manually identifies peaks missed by the peak detection algorithm. Due to the level of suer involvement required, current peak detection is more art than science, and not amenable to high throughput or field deployable applications.