Visualisation Protein intensities on Tissue

[melissaklingeberg]

I would like to visualise the protein intensities from the DIANN output on the tissue image inside the LMD contours.
Is that possible with openDVP and what is the required input? Do you recommend using non-log or log2-intensities for this?
thanks in advance!

[josenimo]

Dear @melissaklingeberg,

thanks for bringing this up.

Regarding visualising proteomics on top of images

this is one of the great advantages of using SpatialData with openDVP.
In tutorial 2 I go through the main motions of what is needed to project the information on your images.

In summary, you need:

1. Proteomics data (DIANN format so far)
2. Shapes used to collect as geojson file
3. Images to overlay on

These files must be connected/link in the following way:

1. Images and shapes must have the same coordinate system, meaning, the shapes must have been annotated from that image.
2. Shapes and proteomic data must share a common sample_id, in order to know which shapes should have which samples.

Consider:

• Plotting very large images (+25gb) can be slow in most laptops and office computers, consider using a workstation, or compressing your image to lower the information density.
• SpatialData uses napari as an interactive visual interface, which is useful but also lacks many features like (coloring samples based on categorical, search a variable in the menu etc.). Spatialdata also offers a matplotlib/ggplot interface called spatialdata_plot which can be quite useful for elaborate visualizations.

Regarding non-log or log2 instensities

Whatever you feel is most intuitive for your public, I would suggest Zscore (variable-wise)