Cell Type annotation showing Normal cells as Cancer cells in single cell analysis.

Question

[geno89]

Hi

I am using Sctype for cell type annotation on the Xenium spatial transcriptomic Bladder tissue sample of mouse bladder

source("https://raw.githubusercontent.com/kris-nader/sp-type/main/sp-type.R")


 xen <- run_sctype(xen, known_tissue_type="Immune system", slot="SCT")

The problem is it is classifying all the urothelial cells as Cancer cells even in Normal Samples too. What could be the problem in this annotation?

[ATpoint]

The annotation is as good as the reference. If it is misclassified then probably the set of reference genes is not good / poor. What this probably does is to calculate some sort of score for similarity with the reference. Can be that there is no good marker set in your reference for normal urothelial cells so it classifies them by "the next best" which could be cancer cells, no matter whether the score is good or not.

[geno89]

Hi Thanks for your answer. Is there anyway to validate this theory? I am just worried if the normal is not normal sample or perhaps mislabeled?

[rfran010]

Immune System seems like the wrong tissue to use. Have you looked into whether bladder is present in the scType database as well?