Closed bschilder closed 3 months ago
bschilder
commented
3 months ago After re-reading the preprint, it seems like the logistic classifiers were only trained on specific cell types for specific use cases. Given that I need to do prediction for all cell types, I've just gone ahead and trained my own classifier. Happy to share if this is helpful.
Still might be nice to share the code for the classifiers bit of the manuscript. But i'll leave that up to you.
yassinho-f
commented
2 months ago After re-reading the preprint, it seems like the logistic classifiers were only trained on specific cell types for specific use cases. Given that I need to do prediction for all cell types, I've just gone ahead and trained my own classifier. Happy to share if this is helpful.
Still might be nice to share the code for the classifiers bit of the manuscript. But i'll leave that up to you.
Hey @bschilder ,
sounds great, I am also interested in training the UCE embeddings of the IMA_sample.h5ad! Would you like to share the code?
Hi @Yanay1 , super excited to use this foundation model! Thanks for making the code and file so easy to use.
In the preprint you trained logistic classifiers on the IMA to predict the cell type identity in new datasets. I don't suppose you'd be able to share one of these pretrained classifiers for predicting cell type identity across all major cell types seen in the IMA? This would make one of the more common tasks of cell type annotation even easier.