Resumen de Development of a complete advanced computational workflow for high-resolution ldi-ms metabolomics imaging data processing and visualization
Mass spectrometry imaging (MSI) is a technique that can map the spatial distribution of an analyte directly onto a tissue section. This allows representing specific molecule distributions directly from a tissue section. MSI is becoming a valuable technology for histopathology since rich chemical information is recorded in a single experimental run. Three main ionization methods have been developed for MSI: MALDI, SIMS and DESI. In this work, MALDI (matrix assisted laser desorption/ionization) is used due to its advantages compared to other ionizations techniques. MALDI provides a high spatial resolution with good ionization characteristics of low-weight compounds. In MALDI, a laser scans the sample surface and promotes the ionization of each pixel in the image. MALDI is an established technique for the acquisition of the high mass range of the MS spectrum. However, it is still not widely adopted for metabolomics studies. The MALDI acquisition of low molecular weight compounds is a challenging task mainly due to the MS signal interferences introduced by the organic matrices compounds used to promote the ionization. We have developed an alternative laser desorption/ionization (LDI) method to improve the MSI detection of metabolites. Our LDI method consists in coating the tissue with a gold nano-layer. This nano-layer is deposited by means of the sputtering technique which is a very robust and repetitive process. In contrast to classic MALDI, no solvent is used to deposit the gold nano-layer since sputtering is a dry deposition procedure. This overcomes the problem of compound lateral diffusion of sprayed MALDI matrices enabling the MSI acquisition at ultra-high lateral resolution. The sputtered gold nano-layer also provides a reliable method for obtaining low mass range MSI datasets because very few background MS signals are generated from the sputtered layer. Moreover, the MS peaks corresponding to gold clusters appear homogeneously distributed throughout the MS spectrum at every image pixel. This enables an accurate mass calibration by using the gold MS peaks as mass references.
The following step after the MSI acquisition is the data processing. MSI generates a large quantity of complex spectral data. Translating the MSI raw data into relevant chemical information is still a challenging task because of such factors as the experimental variation and the huge size of the MSI data. This requires implementing computationally efficient routines to process the raw MSI data. To address this, we developed two software packages for the R platform: rMSI and rMSIproc. These software packages establish a novel and flexible platform for MSI data analysis, completely free and open-source.
The rMSI package is focused on providing an efficient way to manage MSI data together with a graphical user interface (GUI) integrated in R environment. MS data is loaded in rMSI custom format optimized to minimize the memory footprint yet maintaining a fast spectra access. The data format is designed to place all the data in the hard drive following a matrix-like structure. Then, only the data chunks needed at each time are automatically loaded in the computer memory. This allows an appropriate management of larger than memory MSI datasets. The rMSI GUI is designed for simple and effective data exploration and visualization. Moreover, rMSI is designed to be integrated in the R environment through a library of functions that can be used to share MS data across other R packages.
The rMSI package provided us with a solution to manage and visualize MSI large datasets. However, it is necessary to assign MS peaks to chemical entities in order to extract relevant biological information from the MSI experiment. This analyte annotation process is intrinsically linked to the mass accuracy of the data. Mass accuracy and analyte identification are determined by such factors as the experimental set up and the data processing workflow. We present an MSI data processing workflow that uses a label-free approach to compensate for mass shifts. The algorithms developed were designed to perform efficiently even for large datasets generated from an FTICR mass spectrometer. We assessed the overall mass accuracy in the range m/z 400 to 1200 using silver and gold sputtered nanolayers. With our novel processing workflow we were able to obtain mass errors as low as 5 ppm using a TOF instrument. This mass accuracy enhanced workflow is implemented in the rMSIproc package. Besides, rMSIproc also includes a complete preprocessing pipeline able to produce a reduced peak matrix from an MSI experiment performed with TOF or FTICR spectrometers. The generated peak matrix is a data reduced but accurate representation of the whole MSI dataset. Moreover, the peak matrix is also small enough to fit in computer memory. Thus, this enables the use of previously developed statistical analysis algorithms to be easily applied to MSI datasets. rMSIproc takes advantage of rMSI data model to work with files larger than the computer memory capacity. Most of the rMSIproc internal routines are implemented in C++ using a multi-threading strategy. This allows to take profit from modern multi-core processors thus provides a better processing performance to the open-source MSI data analysis.
We believe the developed experimental workflow together with the developed software packages will have a positive impact on MSI for spatial metabolomics applications. In our opinion, this work will contribute to a future better understanding of modern molecular histopathology from the point of view of metabolomics.
