Extracting only tumor samples from Single cell RNASeq

Question

[David_emir]

Hi All,

I am currently working on Singel cell RNAseq Tumor data, and I have around 15000 Tumor cells from 7 cancer patients. I am interested in looking into the expression pattern of genes in only tumour cells. I am sure the 15000 cells which we classified are not 100% tumour cells. So my question is, do we have any methods where we can extract only tumour cells. I tried Infercnv, Conics but I am not able to get only Tumor cell annotations. It would be great if you guys can help me in solving this issue.

Thanks a lot, Stay safe.

DA

Answer 1

[jared.andrews07]

Does your cancer have any known markers? The easiest way is to just look for clusters that express tumor-associated markers. Any very common coding mutations? You could try to look for clusters enriched for those mutations via Vartrix. Are there other single cell transcriptomes for those tumor types published that have malignant/normal cells annotated? If so, you could try labeling via SingleR or other reference-based cell annotation methods. If these are blood cancers, you may be able to look for dominant TCR or V(D)J clonotypes.

[David_emir]

Thank you so much for your help. I have only a matrix of counts (10X) and i am not sure if Vartix can be used on this type of data. can you please guide me in this case? Thanks a lot.

[jared.andrews07]

You need pre-defined variants for Vartrix. Typically it's used in conjuction with WGS/WES from bulk tissue to look for clonal variants in different cell types/tumor populations. If there are very common mutations in your cancer-type, you could try creating a VCF of those mutations, which Vartrix would then look for in the BAM files. Obviously, if you don't have those files, it's not going to be of any use to you.

Are there no none malignant markers in your tumor of interest?

[David_emir]

Thanks a lot, Dr jared.andrews07, I have subsetted out epithelial cells, can you please let me know the malignant marker gene sets? or any reference so that I can check them and group them into Malignant and Non-Malignant. Thanks a lot for your help.

[jared.andrews07]

This depends wildly on your cancer type. You should consult the literature, your PI, or local experts on the cancer in question to determine appropriate malignant cell markers.