Artifact detection with respect to lacking mitochondrial genes in Visium FFPE chemistry #19
Description
First of all, I've enjoyed reading your preprint, interesting work!
You state that
SpotSweeper can be used with spot-based SRTs [...]
such as 10X Visium, and also that
[...] SpotSweeper currently uses multiscale variance of mitochondrial ratio to detect these artifacts, [...]
However, I would like to point out that the Visium Spatial Gene Expression platform comes with two different chemistries, one for fresh-frozen (FF) and one for formalin-fixed paraffin-embedded (FFPE) tissues. The latter does not detect any mitochdonrial genes at all, as you can read in the 10X Q&A for FFPE: What genes will not be detected by the Transcriptome Probe Panels?:
The list [of genes detected by the Transcriptome Probe Panels] was limited further by excluding mitochondrial and ribosomal genes, [...]
Therefore, I think it would be great to add some distinction at the beginning of the vignette which outliers/artifacts can be detected in Visium FF vs. Visium FFPE, i.e. which features of SpotSweeper don't rely on information coming from mitochondrial genes.
As far as I understand the vignette, local outliers could be identified in Visium FFPE (via localOutliers using library size and unique genes) but dryspots and hangnail artifacts couldn't be identified at all (via findArtifacts) because that would require information about mitochondrial genes.