Implémentation d'un programme permettant l'annotation des lipides provenant de données MSI dans le but de segmenter une coupe histologique en des régions métaboliquement différentes

Abstract

Mass Spectrometry Imaging (MSI) is an analytical method employed to map the distribution of molecules inside a sample based on their mass-to-charge ratio. When working with biological sample, classical histology is often combined due to the complementarity of the data. Unfortunately, bioinformatic tools used to treat those data are not open source and are often limited in terms of their analytic possibilities. The lack of means to analyse imagery by mass spectrometry data forces scientists to build their own tools. In this work, we have developed a tool from existing methods to process image by mass spectrometry data in order to identify lipids in an image by MALDI FT-ICR mass spec- trometry from a biological section. The different methods used to achieve this goal are peak picking methods, to process the mass spectra signals, followed by a FDR controlled anno- tion of lipids. The identifications are made by the LIPID MAPS database. Furthermore, not recorded lipids are infered according to the lipids already identified. Moreover, we used the identification process as a feature reduction method to allow efficient segmentation of an image by mass spectrometry. Then, the regions of interest, discovered by the segmentation , can be analyzed in term of their lipid composition. The pipeline is implemented in R and is tested on 3 biological sections of zebrafish.