T cell subtype annotation

Question

[georgii.vdovin]

Hello, I have more of a general question. I am annotating CD4+ / CD8+ T cell subtypes in a non-model organism. What metrics would you use to label a cluster as Cd4+ with some level of confidence? Anything other than non-zero median?

[allingt]

I guess reference-based annotation is not an option? In my experience, visualising the expression of canonical markers (I don't know if CD4 is the only relevant one?) for clusters gives a good overview of the clearest clusters. For example dotplot function in seurat. You can also see in a dotplot what the lowest CD4 expression in any cluster might look like, to help guide choice of some threshold. In addition, you can perform differential expression analysis to find a posteriori markers that distinguish each cluster.