search

dviraran / xCell

Cell types enrichment analysis
182 stars 62 forks source link

Batch effects in input data and reliability of applying xCell to non-human species #51

Closed cwang131 closed 4 years ago

cwang131 commented 4 years ago

Hi,

I've been trying to apply xCell to some data that has batch effect issue across groups of interest. I wonder if aggregated data containing those from different batches without batch correction could be used as input for xCell scoring and if downstream comparison of a cell-type score across groups could still be reliable. The manuscript touched briefly on the robustness against batch effect in the discussion. I'd appreciate your insights.

In addition, related to a previous issue (https://github.com/dviraran/xCell/issues/26), I wonder if xCell could be applied to non-human species, such as rats, mice, and ferrets.

Thank you!

dviraran commented 4 years ago

Hi,

Regarding batch effects - xCell uses gene expression ranks and not the levels themselves. This means that different normalization and batch effect corrections will have little effect on the xCell analysis.

Regarding non-human, I don't recommend it, it is not trained for non-human species. However, I know that some groups got ok results with the current version of xCell in other primates and mice.

Best, Dvir

cwang131 commented 4 years ago

Thank you very much for your help! I really appreciate it.