Using MAGIC with a combined dataset

[saeedfc]

Hi,

I have to combine several datasets. Batch correction is used for clustering and Dimensionality reduction. But for visualization and DEG analysis I will be using normalized RNA counts (Seurat NormalizeData() function). Hence I have two options to get an imputed matrix for gene expression visulaization or DEG analysis.

1. Do magic imputation on individual datasets and combine the matrices.
2. Impute the combined matrix with magic.

What would be best? Or it doesn't make any difference?
Any Help is appreciated,
Kind regards,
Saeed

[Emma1997WTT]

Hello, I am facing the same problem, do you have an exact answer now?

[LiuCanidk]

Hi, the same problem, is there any progress now?

[saeedfc]

MAGIC contributors do not often respond here. They used to have a slack channel which I can't find now.
I thinki: It is likely that imputation of individual batches exacerbates batch effect. But I would try both and see if this is true.

[LiuCanidk]

@saeedfc thanks for your reply!