scVI vs Harmony, which is better for cell type based clustering in brain tissue?

Question

[biotrekker]

Hello, I have a post-mortem dataset with 50+ donors. I do not have a brain region specific reference so I cannot use scANVI. I was wondering whether scVI or Harmony would be better in my case? I know both are good and available in python.

Thanks

[Arup Ghosh]

Try the references from the following publications.

https://www.nature.com/articles/s41591-024-03150-z

https://www.science.org/doi/10.1126/science.add7046

Answer 1

[ATpoint]

There is no "better" is my experience. It always depends on the dataset. Try both and integrate your biological knowledge to decide which makes more sense. I always go with harmony first (and fastMNN from batchelor second) as it is fast and usually sufficient.