Imaging with high-resolution mass spectrometry (MSI) such as Orbitrap mass analyzers yields large datasets that contain the spatial distribution of all compounds that can be sampled from a tissue section into the mass spectrometer. In this talk we present a bioinformatics approach that allows rapid exploration of a large MSI dataset without loss of information, while generating high-quality ion images of compound isotopes in accordance with the resolution of the mass spectra.
Our approach includes peak detection in all collected spectra, construction of a consensus peak list, and generation of ion images of compound isotopologue peaks using the resolution of the mass spectra at given m/z. Scoring the content of the spatial structure by the first principal components and using spatial chaos metric allows culling of ion images with low information content.
Our tool enables spatial correlation queries using ion images, selected area or combination of the selected area and ion distribution. The performance of our approach over regular binning and other tools such as Cardinal is demonstrated by automatic generation of separate clean ion images of spiked-in drug compounds or lipids close in mass. We further demonstrate the tool’s performance by its ability to detect a large number of isotopes and in-source fragments of spiked-in cancer drugs in correlation queries.