Do you recommend this approach to use it for scRNA Seq data?

[biobug16]

Hey @labrazil and @tiagochst

I have scRNA Seq (10X genomics) data from a cancer and want to quantify the stemless indices for each cell type or cluster. Do you recommend this approach for the same?

If yes, then I have another question, should I use the bulk tissue data from stem cells (ES or iPSCs as in your paper) to train the model and use our data for testing or should I use single cell data itself for training the model?

I ll really appreciate any type of insights on this.

Thanks

[tiagochst]

Hello,

Here is the answer from the first author of the paper (Tathiane Malta).

Although we have built our model using bulk cells and applied in bulk tumor samples, we also tested in single-cell data in the paper. Figure 5C shows the prediction model applied to single-cell data of glioma and Fig 5D shows single-cell data of breast cancer.
It is difficult to tell the biological meaning without further validation though. For the glioma samples, our model was positively correlated with an independent stemness score described by the original paper, which is good I think.
In summary, I believe you can use our approach to your single-cell data. The idea of training a model using single-cell data of stem cell looks interesting as well.
I hope this helps.
Thank you for your interest.