A new peak detection algorithm for MALDI mass spectrometry data based on a modified Asymmetric Pseudo-Voigt model

Area of Science:

• Analytical Chemistry
• Biophysics
• Computational Biology

Background:

• Mass spectrometry (MS) is crucial for analyzing bio-molecules by measuring mass-to-charge ratios.
• Accurate peak detection and parameter estimation are vital for identifying bio-molecules and their abundances in MS data.
• Existing methods struggle with overlapping and asymmetric peaks in mass spectra.

Purpose of the Study:

• To develop and validate a novel algorithm for enhanced peak detection and quantification in mass spectrometry data.
• To improve the accuracy of peak parameter estimation, including summit location and area.
• To address limitations in current MS data analysis, particularly for complex spectral features.

Main Methods:

• Utilized dual-tree complex wavelet transformation and Stein's unbiased risk estimator for spectra smoothing.
• Employed a modified Asymmetric Pseudo-Voigt (mAPV) model for peak modeling.
• Integrated hierarchical particle swarm optimization for robust peak parameter estimation.

Main Results:

• The mAPV model demonstrated superior fitting accuracy for asymmetric peaks compared to Gaussian, Lorentz, and Bi-Gaussian models.
• Achieved lower percentage errors in peak summit location (0.17-4.46% less) and peak area estimation (approx. 0.7% less than Bi-Gaussian).
• The overall algorithm showed improved sensitivity (85%) over benchmark methods (77% and 71%) on MALDI-TOF data.

Conclusions:

• The developed algorithm excels in peak detection and parameter estimation for MS data, especially with overlapping and asymmetric peaks.
• The method offers significant improvements in accuracy and sensitivity for bio-molecular analysis.
• The algorithm is implemented in MATLAB, with source code publicly available for broader application.