I'm using scANVI-scArches to integrate data from several studies. Each study has several samples. So I set up the model with categorical_covariate_keys=["sample", "study"]. This worked when integrating the data. However, when I was trying to project new data to the reference data. I got an error saying scArches currently does not support multiple categorical_covariate_keys. Given this situation, is it reasonable only including batch_key="sample", ignoring the differences between studies? I also noticed that HLCA atlas was also built from several studies with multiple samples. How did you deal with this situation?
Hi,
I'm using scANVI-scArches to integrate data from several studies. Each study has several samples. So I set up the model with
categorical_covariate_keys=["sample", "study"]
. This worked when integrating the data. However, when I was trying to project new data to the reference data. I got an error saying scArches currently does not support multiple categorical_covariate_keys. Given this situation, is it reasonable only includingbatch_key="sample"
, ignoring the differences between studies? I also noticed that HLCA atlas was also built from several studies with multiple samples. How did you deal with this situation?Thanks, Yuqi