Closed tinakeshav closed 1 year ago
Hi Tina,
If you have two datasets which are quite similar, your options depend on the strength of batch effects between them. You could train a model on one dataset, save it, then continue the analysis with the other (if the cell compositions/types are similar), but this may be confounded by batch effects. Another strategy might be to model all the samples together. Then, you would at least know the extent of batch effects and it is likely some topics would still be shared between samples.
Let me know if this helps, Allen
Hi! Super excited to play around with MIRA. I have a dataset that might benefit from transfer learning since they are quite similar samples. I was wondering if this is an achievable goal with MIRA and recommendations for how? Thanks a ton! Tina