InferCNV how to choose the right reference?

Question

[Виктория]

Hello everyone! Could you help me to choose the correct reference for scRNA data if I need to identify CNVs among the malignant clusters that are represented as immune cells with an epithelial phenotype (CTC). What is more correct to take healthy immune cells or healthy epithelium as a reference? When I establish a reference group as immune cells, the data obtained do not quite coincide with the signaling pathways and it finds CNVs only in clusters with platelets.

Answer 1

[jared.andrews07]

We're not really going to be able to answer this for you, as we don't know your experiment/samples. I used to just take all healthy cells rather than limiting to one type.

[Виктория]

Thank you! One more question. Is it okay to take healthy cells from external dataset for that comparison?

[jared.andrews07]

You'll get fewer false positives and such if the cells are matched. Any germline changes will be properly captured in matched cells, not so if they're pulled from a different sample/experiment/study.

Answer 2

[xingyu]

I think ... The optimal option is control epi cells from health samples, if you have. If you don't have health control epi cells, the immune cells is good choose. In addition, datasets from differient source have batch effect and it will hinder to find the true outcome.