EasyMSI: a competitive software tool for the interpretation of Mass Spectrometry Imaging datasets
Abstract
EasyMSI is a software tool for spatial and spectral visualization and processing of mass spectrometry imaging, with several modules providing user assistance for the interpretation of data. It has been developed by CEA-LIST during the Computis European project (2006-2009) and the Masda-Eye ANR project (2010-2012). EasyMSI can read MALDI and SIMS datasets in Analyze format (Applied Biosystems), GRD format (IonTof) and the imzML standard format for mass spectrometry imaging based on mzML. EasyMSI provides basic functionalities for data display and exploration (spectrum and image display, peak and pixel picking, zooming on spectra and images, ROI selection), and some more specialized treatments such as denoising spectra or structure analysis. EasyMSI offers the advantage to display MALDI data without binning. The user can choose to
bin or not SIMS data, and the associated binning rate. As assistance to the interpretation of data, EasyMSI computes several indicators. The relative variance spectrum enhances peaks that have a highly contrasted space distribution. The Moran
index is an autocorrelation indicator adapted to detect thin local structures such as membranes, in images. The correlation spectrum associated to a given m/z brings out correlated m/z, often co-localised or complementary with the given m/z.
EasyMSI is able to extract the peak list of the total spectrum, with parameterized criteria of extraction, to best adapt to the dataset peak shapes versus noise. Peak list can be used for molecule identification through the interrogation of biological databases. EasyMSI also includes clustering tools to perform spatial (i.e. pixel-based) classification or spectral (i.e. m/z-based) classification on peak lists or binned data. The K-means clustering is one of the simplest and fastest classification methods. The time for running a K-means clustering usually lasts only some seconds. The hierarchical clustering enables to perform
clustering inside a zone defined by a preceding clustering. The diffusion map method consists in reducing dimensionality by embedding data in a space in which data are more easily synthesized, and to do a clustering analysis in the reduced data.