Deck 15: Technology-Specific Tutorials -> Real Numbers

In the Visium tutorial, what genes are shown as examples of highly-expressed genes to color spots by?

Mitochondrial genes like mt-Co3 and mt-Co1.

How many fields of view does the MIBI-TOF example dataset contain, and what are they called?

3 → point8, point16, point23, each its own coordinate system.

In MIBI-TOF, what are the table's var_names instead of gene names?

Protein markers (e.g. ASCT2, CD11c, SMA).

Roughly how many points does the MERFISH example dataset contain, and how does the tutorial handle plotting them?

Over 3 million; the tutorial subsamples to 1% before plotting.

What is the Visium HD capture area size, and the smallest bin size?

6.5mm × 6.5mm capture area; 2µm smallest bins, giving ~10.5 million bins.

How much memory would a full raster representation of Visium HD data require (per the tutorial's estimate), vs. the actual sparse representation?

~187 GB fully rasterized, vs. ~130 MB for the actual sparse vector+table representation.

What artifact appears when plotting the (slightly rotated) Visium HD grid directly with matplotlib, and what mitigates it?

A Moiré pattern; switching to method="datashader" mitigates it.

What does scale="full" do in render_images(), and why is it essential for viewing rasterized-bins layers?

It disables auto-rasterization to canvas size — essential so the rasterized-bins layer is shown at its actual computed resolution rather than being re-rasterized.

What new capability does the Visium HD SpaceRanger 4.0 tutorial add over the standard Visium HD tutorial?

Cell and nucleus segmentation (not just binned data), read via load_segmentation_only=True, load_nucleus_segmentation=True in the visium_hd() reader.

What function derives point centroids from a Shapes element in the SpaceRanger 4.0 tutorial?

sd.get_centroids()

What are the 4 channels used for cell segmentation in the Xenium Cell-Segmentation-Kit dataset?

DAPI (nuclear), ATP1A1/CD45/E-Cadherin (boundary), 18S (interior-RNA), AlphaSMA/Vimentin (interior-protein)

How can you overlay the same gene expression on a different Shapes element (e.g. circles vs. full cell boundaries) in Xenium data?

Reassign the table's annotation target with sdata.set_table_annotates_spatialelement("table", region="cell_boundaries")

What is SpaceM, and what does its table's var_names represent?

A spatial metabolomics method combining MALDI imaging mass spectrometry with microscopy; var_names are ion names, not genes.

Why can SpaceM's expression matrix contain NaN values?

Because the dataset integrates many METASPACE datasets, and an ion may be detected in some but not others.