Some questions about infercna

[YiweiNiu]

Dear Laffy,

I read your new glioblastoma paper and would like to try your methods on our single-cell RNA-seq data.

But I have the following concerns.

1. What is the differences between this package and infercnv? Intuitively, I found infercnv is much more complicated.

2. When using infercna(), can I pool all samples (different tumors) together? Or it should be applied to individual sample? Since Figure 1A in your paper show all samples in one heatmap, I guess you put all cells in one matrix, is that true? I do not know if pooling different tumors is a good way, as inter-tumor heterogeneity may influence the results.

3. Did you use refCells in analyzing the glioblastoma data? Or you classify malignant cells and non-malignant cells by combining CNA signal and CNA correlation? In the method "CNA inference from single-cell data", it said cells were classified by CNA signal and CNA correlation, but the upper panel in Figure 1A shows the CNA levels of non-malignant cells. So, I am confused.

It would be highly appreciated if you could give me some hints. I am new to single-cell data analysis.

Best wishes,
Yiwei Niu

[twoneu]

Hi @jlaffy,

I am also curious about @YiweiNiu's questions 1 and 2. Do we expect the results to be comparable between this package and inferCNV?

Also, what do you recommend for a sample containing few/no normal reference cells?