A Novel Peak-Shape Aware Approach for Mass Alignment in Mass Spectrometry

In mass spectrometry measurements, mass shifts may be inadvertently introduced due to instrumental drift and calibration inaccuracies, potentially compromising the accuracy of subsequent data analysis. This work presents a novel, label-free algorithm to improve relative mass alignment between mass spectra. The warping function is modeled as a natural cubic spline, a suitable model for gradual, nonlinear mass shifts.

The algorithm is validated on a dataset generated from human glioblastoma multiforme samples using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry imaging (MALDI-TOF MSI) and rapid evaporative ionization mass spectrometry (REIMS), and public datasets generated from MALDI-TOF and desorption electrospray ionization (DESI) Orbitrap instruments.

The algorithm considerably reduces the mass dispersion of the dataset and improves the similarity to a reference mass spectrum for MALDI-TOF, REIMS and DESI-Orbitrap data. It is demonstrated that the proposed method reliably corrects for severe mass shifts. The algorithm is competitive in speed and mass dispersion reduction compared to MSIWarp, a common method to correct for mass misalignment.

This paper presents a novel algorithm to reduce relative mass misalignment for mass spectrometry, validated on an extensive dataset. Thanks to the use of profile data, the peak shapes contribute to the computation of the warping function, and the warping function is approximated in a robust manner. An open-source Python implementation of the proposed methodology will be made accessible on GitHub: https://github.com/VanhemelThomas/psalign.