xCompass-AI / GeneCompass

GeneCompass
https://xcompass-ai.github.io/GeneCompass/
37 stars 3 forks source link

Questions about preprocessing #4

Open HelloWorldLTY opened 1 month ago

HelloWorldLTY commented 1 month ago

Hi, I notice that you have an example for pre-processing datasets, however, it seems that this code does not provide information for the approach of including metadata (cell type, dosage information) or splitting train/test dataset. Would you please share more information about how to use your model for our own dataset?

https://github.com/xCompass-AI/GeneCompass/blob/main/preprocess/preprocess.py

HelloWorldLTY commented 4 days ago

Moreover, I wonder that what is the difference between the preprocessed codes used here and the preprocessed codes used in Geneformer? It seems that they are both transferred with ranks. Can I directly use the codes from Geneformer to test this model? Thanks.