Comparison of algorithms for profile-based alignment of low resolution MALDI-ToF spectra

Abstract

Recently MALDI-ToF mass spectrometry has become major technique for the analysis of complex protein mixtures. For large studies or when data from different machines are compared, a problem of shifts of peaks between spectra becomes very urgent, especially in low resolution data. In our study we compare different profile-based alignment algorithms using a virtual instrument.We investigate the performance of algorithms through different simulation scenarios and indicate their weaknesses and strengths by a few commonly used statistical measures. We propose a modification of the peak alignment by fast Fourier transform algorithm, which significantly improved its robustness by reducing alignment artefacts and over alignments. Spectral alignment is an indispensable element in the MS data pre-processing. PAFFT algorithm with our modification may be a good MS spectra pre-processing tool, by achieving high sensitivity and keeping false discovery rate at a reasonably low level.