DeisoLAB - isotopic envelope identification by analysis of the spatial distribution of peptides in MALDI-MSI data

Le résumé fourni par la source

A new approach (DeisoLAB) is introduced to handle extensive peptide data obtained in mass spectrometry imaging experiments and detect potential isotopic envelopes. It is based on a combination of a fuzzy-inference system and analysis of spatial distribution of peptides. Potential isotopic envelope members are identified in the first step using the Mamdani-Assilan fuzzy-inference system. Then, preselected isotopic envelope members are analyzed in the second step regarding their spatial distribution. The spatial distribution of an analyte is visualized as a spatial map of molecular distribution that reflects the peak intensities registered for each m/z across the whole tissue section. When comparing the spatial maps of molecular distributions of different isotopic envelope members, it can be observed that there is a difference between those included in one isotopic envelope and those not included. To measure this difference, several image texture metrics are applied. Based on those metrics, using the Naïve Bayes classifier, peaks can be classified into two classes, envelope and non-envelope. The method was evaluated on eight MALDI-MSI datasets, including fresh-frozen and formalin-fixed paraffin-embedded tissues. The fuzzy-inference-system-based filtering reduced the number of candidate peak pairs requiring further analysis, while the subsequent spatial-distribution-based classification achieved recall values of 88.12–94.12%, precision values of 72.95–85.71%, and specificity above 99%. The overall deisotoping accuracy reached 96.98%. Currently, several tools are already available enabling analysis of peptide envelopes, however they use predominantly spectral information, whereas we propose a computational method (DeisoLAB) that additionally exploits spatial molecular distributions of species – a feature unique to MSI data. This allows for substantial broadening of the possibility of differentiation between individual isotopic envelopes, also in the case of mass spectrometry data of relatively low mass resolution.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.