annotation of brain cells using pretrained brain model

[surPoudel]

Hi,
This is a great tool and I am trying to implement it in mouse brain data. I have downloaded the model using the link. I wanted to predict the cell types in my dataset using the pretrained model and this should be pretty simple as I am not trying to use reference to query prediction. Is there anything in tutorial that I can leverage to do this simple task? Also can we preprocess and normalize the data ourselves and just use scGPT model to predict the cell types?

Thank you.

[subercui]

Hi, thank you for the question. Could you provide more details about your use case? To clarify, the model does not store any cell type names or annotations internally, so there may need to be some "reference" provided in a sense. (1) If you have or know a dataset that has labels, you may train a classifier on the dataset and use that to annotate your new data. (2) We do have a tutorial about using the "default" cellxgene data and labels and mapping these onto new data. That can a simpler choice for you as well. Please let us know your use case.

BTW, I think this will also need a mouse model, which is indeed a potential enhancement we have been working on.