New use case: Mass spectrometry imaging (MSI)

[foellmelanie]

Hi all,

I made a pull request with three exemplary annotations for MSI data to start a discussion: #450

Here are some thoughts about the annotation of MSI data with the current sdrf file format. Especially the 'MSI specific things' would benefit from a discussion.

Metadata that applies to MSI as well

• Label: label-free is default
• Cleavage agent details: Trypsin is default, but also other analytes (lipids, glycans, proteins) are imaged and "not applicable" currently fails the test
• Instrument (Bruker rapifleX and timsTOF fleX are missing in current Ontology)
• Characteristics of sample such as organism, age, disease,…
• Spiked compound: Internal calibrants are common in MSI, as MSI does not apply fasta files/MS2 searches it would be helpful to be able to add modified peptide sequences and/or the theoretical mass of the compound.

Metadata that does not apply to MSI

• Reduction and Alkylation: Was suggested to be mandatory. In that case for most MSI studies it would be “not applicable”
• Fraction identifier: this is never done in MSI → Therefore, I would prefer “not applicable” in contrast to classical unfractionated proteomics experiments, where one would set all fractions to “1”
• MS2/Database search related terms are not mandatory currently, therefore they could just be ignored for MSI data.

MSI specific things

• File type: The open standard format imzML is a composite file type consisting of two files: imzML + ibd
  → In the current annotation I only used one line and referenced only the imzML subfile but not ibd. This makes it easier to annotate and understand the metadata but might be problematic for some (automated) applications?

• Multiple samples per run: Multiple tissues can be placed onto one slide As the technique is label-free and spatially resolved the samples can be distinguished by their position on the slide/plate. This might require an additional graphic that shows an ID of each sample or the center of the sample in x and y coordinates.
  All files in PXD011104 have multiple samples, I have picked only one of the files to annotate a first simple example. I just numbered the 4 tissue sections (largest image shown here: ftp://ftp.pride.ebi.ac.uk/pride/data/archive/2020/08/PXD011104/Information-gastricmousemodel-MALDI-TOF.docx) from left to right as “position on slide”: 1,2,3,4. This would require an additional image that shows the position IDs.

(Potential) additional MSI metadata columns

• Origin of cleavage agent: Autolysis peptides of proteases are important for calibration/QC/contaminant removal in MSI, as they differ e.g. bovine vs. porcine it would be helpful to include this information
• Matrix: For MALDI experiments the name/chemical composition of the matrix is helpful to remove the corresponding peaks during analysis.
• Ionization source: It is quite common to customize ionization sources, then the available “instrument” terms do not reflect the truth anymore. Adding the Ionization source as separate column might be necessary.
• Analyte: One of the MSI datasets I annotated measured phospholipids, another measured N-glycans.

Cheers,
Melanie

[theodev]

Copying here from Twitter discussion:

Here is our minimal metadata from http://metaspace2020.eu/: https://github.com/metaspace2020/metaspace/tree/master/metaspace/metadata, ironed out through the experience of working with 100+ labs, keeping the balance b/w "the more metadata the better" (it's not) & "minimum needed for processing". Also synced w/ https://www.ebi.ac.uk/metabolights/

[ypriverol]

Thanks, @theodev for the comment.