search

jaydu1 / scVAEIT

Variational autoencoder for single-cell integration and transfer learning.
MIT License
6 stars 0 forks source link

scVAEIT for Mosaic Integration without global Shared Features #9

Closed pondandmoon closed 1 week ago

pondandmoon commented 1 month ago

Hi,

I'm very interested in the scVAEIT method and would like to ask if it is suitable for mosaic integration tasks where there are no globally shared features across batches. For example, in one batch, we have scRNA-seq + scATAC-seq data, while in the second batch we only have scRNA-seq data, and in the third batch we only have scATAC-seq data.

Would scVAEIT still be applicable in this case?

Thank you in advance for your insights!

jaydu1 commented 1 month ago

Hi, thanks for your interest, and sorry for the late reply. Yes, it is definitely suitable for the setting you describe. In that case, the model can learn to translate between the two modalities from the joint measurement. However, caution should be warranted if the joint measurements (scRNA-seq + scATAC-seq) are of small sample sizes.

pondandmoon commented 1 week ago

Hi jaydu1,

Thank you for your explanation. And look forward to your more outstanding work.