search

immunogenomics / starCAT

Implements *CellAnnotator (aka *CAT/starCAT), annotating scRNA-Seq with predefined gene expression programs
MIT License
18 stars 3 forks source link

Annotating after integration #5

Closed Jon-bioinfo closed 3 months ago

Jon-bioinfo commented 3 months ago

If the input data to annotate are from multiple expermients/conditions/etc does that affect the results i.e. would you ideally annotate before combining, or no effect?

Also I had to pip install requests after pip install starcatpy in a fresh conda env. Could be added to the dependencies.

dylkot commented 3 months ago

Hi @Jon-bioinfo, thanks for using starCAT!

The results should be pretty robust to data from multiple experiments/conditions. Many artifact variables are incorporated in the T-cell GEP catalog (e.g. the intermediate early genes or the MALAT1 signal) in which case, cells from the different conditions might have higher use of one of those artifact GEPs. But the loadings of the other non-artifact GEPs are usually pretty robust. You might double check if there are any systematic differences in their usage across conditions that seem sketchy to you. Batch signals that are orthogonal to the GEPs should not impact the results.

Thank you, yes I will add requests to the dependencies