Poster (Scientific congresses and symposiums)
Automatic exploration of images using ultra high resolution MS
La Rocca, Raphaël; Tiquet, Mathieu; Far, Johann et al.
2019BSMS/NVMS
 

Files


Full Text
LaRoccaRaphael-BSMS_final.pdf
Publisher postprint (1.25 MB)
Request a copy

All documents in ORBi are protected by a user license.

Send to



Details



Abstract :
[en] Mass Spectrometry Imaging (MSI) is an analytical method allowing the mapping of molecules in thin sections of samples after their ionization/desorption thanks to an appropriate method. In the case of biological samples, the attribution of molecular formula, and if possible structures, related to the m/z values with their localization can be useful for improving our understanding of biological metabolic pathways. Unfortunately, bioinformatics tools dedicated to the analysis of such data are not open source and are still limited in terms of features. These lacks constrained scientists to build their own tools. In this context, we have developed a tool from existing methods to process images acquired by MALDI FT-ICR-MS, aiming the unambiguous identification of lipids. The image is processed automatically by taking into account the spatial distribution of each m/z value (ionic image). The goal of this approach is to discriminate unstructured ionic images from informative ionic images which revealed regions of interest. This classification is made by an automated evaluation of the level of structure within each image. Briefly, the level of structure of an ionic image is determined by counting the number of clusters of pixels displaying similar levels of intensity, in contrast to ubiquitous ions images in which the levels of intensity in the pixels are randomly spread. The MS image is analysed as follows: m/z values corresponding to lipids are selected based on their exact masses provided by FT-ICR MS (similarly to a peak list). These selected m/z are then curated by evaluating their level of structure. The used of a false discovery rate (FDR) to control this latter step is investigated. Finally, the different validated m/z values are exploited to discover regions of interest in the sample by the means of unsupervised learning algorithms. We describe in this work how to compute the FDR and the spatial distribution for each m/z value. The identifications of lipids were performed using the LIPID MAPS database and different informatics tools were developed in the programming language R.
Research center :
MolSys - Molecular Systems - ULiège
Disciplines :
Chemistry
Author, co-author :
La Rocca, Raphaël  ;  Université de Liège - ULiège > Département de chimie (sciences) > Laboratoire de spectrométrie de masse (L.S.M.)
Tiquet, Mathieu ;  Université de Liège - ULiège > Département de chimie (sciences) > Chimie analytique inorganique
Far, Johann  ;  Université de Liège - ULiège > Département de chimie (sciences) > Chimie analytique inorganique
Quinton, Loïc  ;  Université de Liège - ULiège > Département de chimie (sciences) > Chimie biologique
De Pauw, Edwin  ;  Université de Liège - ULiège > Département de chimie (sciences) > Chimie analytique inorganique
Language :
English
Title :
Automatic exploration of images using ultra high resolution MS
Publication date :
2019
Event name :
BSMS/NVMS
Event place :
Netherlands
Event date :
From 31-03-2019 to 02-03-2019
Audience :
International
Name of the research project :
Eurlipids
Available on ORBi :
since 30 May 2019

Statistics


Number of views
108 (25 by ULiège)
Number of downloads
14 (14 by ULiège)

Bibliography


Similar publications



Contact ORBi