Can I use the SpaCET for uncancerous tissues with the default parameters?

[DingtaoHu]

Dear author,
Thanks fro providing such useful and convenient pakcage for stRNA analysis!
I wonder whether I can use the SpaCET for uncancerous tissues (eg. normal sample, liver cirrohosis, and adjacent cancer tissues) stRNA analysis with the default parameters for celltype defination?

[beibeiru]

Hi, Dingtao,

Thank you for your interest in SpaCET.

For uncancerous tissues, I recommend using SpaCET.deconvolution.matched.scRNAseq, which requires a scRNAseq dataset with all possible cell types. Here is the tutorial.
https://data2intelligence.github.io/SpaCET/articles/oldST_PDAC.html

If you use SpaCET.deconvolution with in-house reference against uncancerous tissues, normal cell types (i.e., normal epithelial cells) would be defined as unknown cells, and also probably be mislabeled as fibroblasts or even malignant cells. This is because 1) the in-house reference has limited cell types; 2) the expression profile of some normal cell types might be close to a part of cell types from the in-house reference.

Best,
Beibei

[beibeiru]

Hi, Dingtao,

We recently upgraded our package, which can deconvolve normal tissue.
Setting adjacentNormal = TRUE will skip the malignant cell prediction in our package.

SpaCET_obj <- SpaCET.deconvolution(
    SpaCET_obj,
    cancerType="LIHC",
    adjacentNormal = TRUE,
    coreNo = 12
  )

BTW, we also added the cell reference of hepatocytes and cholangiocytes so that SpaCET can infer their fractions.

Best,
Beibei