USEFUL RESOURCES FOR OUR PRODUCTS

WHAT DO YOU WANT TO DO?

SCILS LAB VIDEO TUTORIALS

SCILS LAB VIDEO TUTORIALS

Examples of Data Analysis Workflows

This page offers a “learning-by-doing” overview of SCiLS Lab and describes typical use cases based on SCiLS Lab.

A short scientific overview of each use case is provided in the SCiLS Application Notes.

The MALDI-imaging data set of a human colon tumor specimen provided by Professor Axel Walch from the Institute of Pathology at Helmholtz Zentrum München, Germany, was the basis of this page and it can be downloaded here.

Because it is impossible to cover the full functionality of the SCiLS Lab software in a “learning-by-doing” course, users may need to consult the software manual for more detailed information.

Note: The videos are recorded with SCiLS Lab, Version 2015, but are also applicable for the recent version SCiLS Lab, Version 2018.

1. Getting Started with SCiLS Lab

This “Getting Started” guide shows how to visualize spectra and m/z images in SCiLS Lab.

SCiLS Lab uses a proprietary file format that can handle data sets of any size. Therefore, the first step in using SCiLS Lab is to convert data from the flexImaging format to the SCiLS Lab format using the flexImaging data importer. The data can then be analyzed using SCiLS Lab.

2. Discovery of m/z-markers co-localized with annotated regions.

In histopathology, MALDI imaging is used to detect markers discriminating a particular tissue type or morphological structure. Typically, such structures are annotated by an experienced histologist based on a visual inspection of histologically stained sections. For further details, see SCiLS Application Note 2. In this chapter, we describe how SCiLS Lab can be used to automatically elucidate m/z markers that are exclusively co-localized with these annotations. It is assumed that your MALDI imaging data acquired has been successfully imported to the SCiLS Lab data format file. After importing the data, the SCiLS Lab file containing the MALDI imaging data and all annotations done in flexImaging.

The first step of MALDI imaging analysis is preprocessing of MALDI spectra. A typical mass spectrum contains a baseline that is inherent to sample preparation and the MALDI measurement. This means that the baseline does not contain any useful information about abundant ionizable molecular compounds present in the sample. Therefore, baseline correction is a standard method of preprocessing mass spectra. In addition, a MALDI imaging data set can be considered as a collection of spectra that have been measured independently, and therefore normalization of spectra is an important task of preprocessing.

In SCiLS Lab 2015a, preprocessing of MALDI MS spectra is performed automatically during import of data.

The goal of this step is to find m/z markers that are co-localized with an annotated region. The following video shows how this step is performed in SCiLS Lab.

The next video showes how correlated m/z images can be visualized. To combine the information of optical images and the m/z images, the opacity of the m/z image can be interactively set using a slider. Moreover, the intensities below and above certain thresholds can be made transparent (invisible). The following video shows how m/z images can be overlayed with optical image modalities.

The TileView feature is a convenient way for a visual comparison of multiple m/z images, in particular for comparing correlated and anti-correlated m/z values.

Annotations can either be imported from flexImaging or regions can be annotated directly in SCiLS Lab. Using this tool, particular tissue types or morphological structures identified by inspection of the optical image can be saved as a Region object for further analysis. Such analysis may include elucidation of co-localized m/z markers as described in the preceding videos or calculating the mean spectrum of the newly annotated Region and comparing it with the mean spectrum of its complementary Region. A Region can also act as a group for the supervised classification.

3. SPATIAL SEGMENTATION WITH EDGE-PRESERVING IMAGE DENOISING.

A state-of-the-art MALDI imaging dataset comprises a huge number of spectra, where each individual spectrum represents intensities measured at ten thousands of m/z bins. This makes manual evaluation unfeasible and interpreting such large data sets requires computational data mining strategies.

Automatic spatial segmentation can be used as a first step in data mining, providing an overview of the dataset and allowing quick detection of prominent features. In this approach, similarities between spectra are statistically determined, and similar spectra are grouped into a cluster. All spectra within a particular cluster are assigned a selected color and displayed as a spatial segmentation map in which pixels are color-coded according to their cluster assignment. For further detail see SCiLS-Application-note3

The segmentation pipeline combines the peak selection, peak alignment, spatial denoising and segmentation. The pipeline can be executed after opening a SCiLS Lab file.

The segmentation pipeline generates a number of new objects. The following video shows how the segmentation map can be explored interactively.

Co-localized m/z values of a certain segment can be found in the same way as in for a Region, as it is described in video 2.2 above.

4. COMPARATIVE ANALYSIS FOR UNCOVERING DISCRIMINATIVE M/Z MARKERS

MALDI-imaging can help reveal candidates for disease biomarkers by comparing tissue samples corresponding to different conditions. This can be achieved by relating the spatial distribution of ions with the histological annotation of the tissue. SCiLS Lab can be used to find m/z-signals discriminating different biological states and thus facilitate the discovery of novel biomarker candidates. For further details see SCiLS Application Note 5.

The discrimination quality of all m/z values is evaluated by Receiver Operating Characteristic (ROC). The ROC is calculated based on the statistical specificity and sensitivity when the intensity of a single m/z value represents the discrimination rule. The ROC is an exploration of what happens to the specificity and sensitivity when different intensity thresholds are applied to the m/z values.

When your data analysis is finished, you may want to export the results in tabular form. The SCiLS Report Table allows you to quickly compress information about peak lists, such as mean spectra for regions or the discrimination power of the previously used tool. The information compiled in the table can be used in other software, such as Microsoft® Excel®.

More workflows coming soon...

APPLICATION NOTES

SCILS APPLICATION NOTE #6:

SCiLS Lab 2D: Spatial segmentation of MALDI FT-ICR MS imaging datasets.

Because of the high mass accuracy and mass resolving power obtainable, the use of an FT-ICR mass analyzer for MALDI imaging is gaining popularity. However, FT-ICR MALDI imaging datasets are typically large in size and mining the data in an appropriate timeframe is a serious challenge. The spatial segmentation approach is one possibility for tackling this obstacle in FT-ICR MALDI imaging. Here we demonstrate the temporal and spatial lipidomic analysis of head and neck tumor tissue.

SCiLS-Application-note6

SCILS APPLICATION NOTE #5:

SCiLS Lab 2D: Comparative Analysis for Uncovering Discriminative m/z-markers.

MALDI imaging is a tool of choice for discovering clinically relevant biomarkers. In this application note, we used SCiLS Lab to analyze MALDI imaging data from skeletal muscle sections with the aim of finding m/z-markers discriminating pathophysiological regions (trauma; trauma adjacent; healthy) in injured skeletal muscle.

SCiLS-Application-note5

SCILS APPLICATION NOTE #4:

SCiLS Lab 2D: Quantitative Measure for Co-localization of m/z-images.

By revealing molecules specifically localized to a particular feature or structure of the sample, MALDI imaging is a powerful discovery tool. The search for such molecules can be approached by identifying a template molecular ion with a distribution specific to the area of interest and then searching for other m/z-values co-localized with the template ion. Using an application from photolithographic structuring, we show how SCiLS Lab can be used to quantify co-localization of m/z-values in MALDI imaging.

SCiLS-Application-note4

SCILS APPLICATION NOTE #3:

SCiLS Lab 2D: Spatial segmentation with edge-preserving image denoising.

Presently, data mining of a MALDI imaging dataset — which consists of thousands of individual m/z-values — is mostly done manually and represents a serious bottleneck in the data analysis workflow. In SCiLS Lab, automatic spatial segmentation can be used as a first step of data mining, providing an overview of the dataset and allowing quick detection of prominent features.

SCiLS-Application-note3

SCILS APPLICATION NOTE #2:

SCiLS Lab 2D: Discovery of m/z-markers co-localized with annotated regions.

In histopathology, MALDI imaging is used as a tool to detect markers discriminating a particular tissue type or morphological structure. Typically, such structures are highlighted by an experienced histologist based on a visual inspection of histologically stained sections. Here, we show how SCiLS Lab can be used to automatically elucidate m/z-markers exclusively co-localized with these annotations.

SCiLS-Application-note2

SCILS LAB VIDEO TUTORIALS (2014 version only)

Examples of Data Analysis Workflows

This page offers a “learning-by-doing” overview of SCiLS Lab and describes typical use cases based on SCiLS Lab version 2014b.

A short scientific overview of each use case is provided in the SCiLS Application Notes.

The MALDI-imaging data set of a human colon tumor specimen provided by Professor Axel Walch from the Institute of Pathology at Helmholtz Zentrum München, Germany, was the basis of this page and it can be downloaded here.

Because it is impossible to cover the full functionality of the SCiLS Lab software in a “learning-by-doing” course, users may need to consult the software manual for more detailed information.

1. Getting Started with SCiLS Lab

This “Getting Started” guide shows how to visualize spectra and m/z images in SCiLS Lab.

2. Discovery of m/z-markers co-localized with annotated regions

In histopathology, MALDI imaging is used to detect markers discriminating a particular tissue type or morphological structure. Typically, such structures are annotated by an experienced histologist based on a visual inspection of histologically stained sections. For further details, see SCiLS-Application-note2 In this chapter, we describe how SCiLS Lab can be used to automatically elucidate m/z markers that are exclusively co-localized with these annotations. It is assumed that your MALDI imaging data acquired has been successfully imported to the SCiLS Lab data format file. After importing the data, the SCiLS Lab file containing the MALDI imaging data and all annotations done in flexImaging.

The first step of MALDI Imaging analysis is preprocessing of MALDI spectra. A typical mass spectrum contains a baseline that is inherent to sample preparation and the MALDI measurement. This means that the baseline does not contain any useful information about abundant ionizable molecular compounds present in the sample. Therefore, baseline correction is a standard method of preprocessing mass spectra. In addition, a MALDI imaging data set can be considered as a collection of spectra that have been measured independently, and therefore normalization of spectra is an important task of preprocessing.

The following video shows how data loading and spectra preprocessing is performed in SCiLS Lab.

The goal of this step is to find m/z markers that are co-localized with an annotated region.
The following video shows how this step is performed in SCiLS Lab.

The last video showed how correlated m/z images can be visualized.
To combine the information of optical images and the m/z images, the opacity of the m/z image can be interactively set using a slider.
Moreover, the intensities below and above certain thresholds can be made transparent (invisible).
The following video shows how m/z images can be overlayed with optical image modalities.

The TileView feature is a convenient way for a visual comparison of multiple m/z images,
in particular for comparing correlated and anti-correlated m/z values.

2.5 Annotating polygonal regions in SCiLS Lab

Annotations can either be imported from flexImaging or regions can be annotated directly in SCiLS Lab.
Using this tool, particular tissue types or morphological structures identified by inspection of the optical image can be saved as a Region object for further analysis.
Such analysis may include elucidation of co-localized m/z markers as described in the preceding videos or calculating the mean spectrum of the newly annotated Region and comparing it with the mean spectrum of its complementary Region.
A Region can also act as a group for the supervised classification.

3. Spatial segmentation with edge-preserving image denoising

A state-of-the-art MALDI imaging dataset comprises a huge number of spectra, where each individual spectrum represents intensities measured at ten thousands of m/z bins. This makes manual evaluation unfeasible and interpreting such large data sets requires computational data mining strategies.

Automatic spatial segmentation can be used as a first step in data mining, providing an overview of the dataset and allowing quick detection of prominent features. In this approach, similarities between spectra are statistically determined, and similar spectra are grouped into a cluster. All spectra within a particular cluster are assigned a selected color and displayed as a spatial segmentation map in which pixels are color-coded according to their cluster assignment. For further detail see SCiLS-Application-note3

The segmentation pipeline combines the preprocessing pipeline, peak selection, peak alignment, spatial denoising and segmentation. The pipeline can be executed after opening a SCiLS Lab file.

The segmentation pipeline generates a number of new objects that are displayed in the lower part of the Tree. The following video shows how the segmentation map can be explored interactively.

Co-localized m/z values of a certain segment can be found in the same way as in for a Region, as it is described in video 2.2 above.

EXAMPLE DATA SETS

A MALDI-imaging data set of a human colon tumor specimen is provided in the SCiLS Lab data format by Professor Axel Walch from the Institute of Pathology at Helmholtz Zentrum München, Germany. The data set reveals distinct tissue types present in the section: the tumor, the mucosa, the submucosa, the muscularis propria, and the subserosa.

The colon tissue sample was provided as formalin-fixed paraffin-embedded (FFPE) sample, MALDI-MS measurement have been performed after in situ enzymatic trypsin digestion of proteins. Altogether, 12,049 spectra were acquired with a lateral resolution of 150 μm, each spectrum covering a mass range of 600-4,000 m/z. After MALDI-MS experiments, the same section were hematoxylin and eosin (H&E) stained and microscopic images have been superimposed with the MALDI-MS data. For more details on tissue samples and data acquisition we refer to T. Alexandrov, S. Meding et al., (Super-resolution segmentation of imaging mass spectrometry data: Solving the issue of low lateral resolution. Journal of proteomics, 2011) and to the SCiLS-Application-note2.

We gratefully thank Dr. Stephan Meding, Dr. Benjamin Balluff, and Prof. Dr. Axel Walch for providing us this data set for demo purposes.

Download example data set:

colon demo data

A MALDI-imaging data set of a rat testis sample measured on an high-resolution FT-ICR MS is provided by Dr. Corinna Henkel and Dr. Arndt Asperger from Bruker. The data is available in the SCiLS Lab peak list format. It was acquired with a lateral resolution of 10 µm and it contains 5,670 spectra in the mass range of 150-3,000 m/z, each spectrum containing 2,951,625 m/z-bins. We thank Bruker for providing us this data set for demo purposes.

Download example data set:

mouse pancreas data set

A MALDI-imaging data set of injured skeletal muscles of the rat is provided in the SCiLS Lab data format by Oliver Klein from Charite – Universitätsmedizin Berlin, Germany. The data set comprises eight muscle specimens containing primary trauma, trauma adjacent, and intact muscles regions.

The traumatized muscle sample was provided as formalin-fixed paraffin-embedded (FFPE) sample, MALDI-MS measurement have been performed after in situ enzymatic trypsin digestion of proteins. Altogether, 36,926 spectra were acquired with a lateral resolution of 80 μm, each spectrum covering a mass range of 800-3,500 m/z. For more details on tissue samples and data acquisition we refer to O. Klein et al., (MALDI imaging mass spectrometry: Discrimination of pathophysiological regions in traumatized skeletal muscle by characteristic peptide signatures. Proteomics, 2014.) and to the SCiLS-Application-note5.

We gratefully thank Oliver Klein for providing us this data set for demo purposes.

Download example data set:

Klein 2015a

A MALDI imaging data set containing a cohort of six sections of mouse pancreas is provided in the SCiLS Lab data format by Prof. Dr. Kathrin Maedler and Dr. Janina Oetjen from the MALDI Imaging Lab, University of Bremen.

This data set comprises of three sections each from two conditions: Mouse pancreas of mice fed with a high fat/ high sucrose diet (HFD) and from normal diet control mice. The sections were covered with sinapinic acid as matrix and data was acquired in the mass range of 2-20 kDa with a lateral resolution of 75 µm. It contains 72,872 spectra with 7,679 datapoints per spectrum.

We gratefully thank Prof. Dr. Kathrin Maedler and Dr. Janina Oetjen from the MALDI Imaging Lab, University of Bremen, for providing us this data set for demo purposes.

Download example data set:

mouse pancreas data set

A MALDI imaging data set of a 10 µm tissue section of a rat kidney is provided in the SCiLS Lab data format. The data set was acquired with a lateral resolution of 200 µm using sinapinic acid as matrix. It contains 3,530 spectra in the mass range of 2-20 kDa, each spectrum containing 7,680 m/z-bins.

We thank Dr. Lena Hauberg-Lotte and Dr. Janina Oetjen from the MALDI Imaging Lab, University of Bremen, for providing us this data set for demo purposes.

Download example data set:

demokidney