Conflict with variable genes and model training

Teichlab / celltypist

A tool for semi-automatic cell type classification

https://www.celltypist.org/

MIT License

284 stars 45 forks source link

Closed mousepixels closed 5 months ago

mousepixels commented 5 months ago

Love your tool!

I am coming across one issue with celltypist.train:

You check if a cell is normed to 10k counts with (np.abs(np.expm1(indata[0]).sum()-10000) > 1) But, if you filter out variable genes, for example, this will always be false. And filtering out features is a recommended step in your tutorials: https://colab.research.google.com/github/Teichlab/celltypist/blob/main/docs/notebook/celltypist_tutorial_cv.ipynb#scrollTo=bearing-product

Thank you!

mousepixels commented 5 months ago

nvm i found the check_expression argument. Carry on!