How to integrated the scATAC sample with scRNA

[millersan]

Hi Tim, Thanks a lot for developing this very useful package. I feel confused with the integrated scATAC samples to integrate with scRNA samples. Here are my steps to integrate the scATAC and scRNA, before the scATAC samples integrated.

####################################### I: Follow "scATAC-seq data integration", to integrated two samples into one. I get the result with 3 assays: "sciPeaks", "peaks" and "integrated"

II: Set the Defultassay(sample) to "sciPeaks" because when I use the Defultassay as "integrated", I can't make the active gene matrix.

III: Make the active gene matrix and create it to sample[["RNA"]]

IV: Remaining steps follow "Integrating with scRNA-seq data" in "Analyzing PBMC scATAC-seq" #######################################

Is there anything wrong with my steps?

And I feel confused with the slot "integrated"

class(integrated@assays$sciPeaks)
[1]` "ChromatinAssay"
attr(,"package")
[1] "Signac"

class(integrated@assays$integrated)
[1] "Assay"
attr(,"package")
[1] "Seurat"

How to uesd the integrated slot?

Thank you so much!

[timoast]

If your goal is to transfer cell type labels, there's no point running the scATAC-seq batch correction as this isn't altering the gene activity values.

I'd suggest that you:

1. Follow the merge vignette to create a merged scATAC-seq object with the same set of peaks quantified in all cells
2. Quantify gene activities using the GeneActivity function
3. Follow the workflow shown in other vignettes to transfer cell type labels from a scRNA-seq dataset.

The reason you can't call GeneActivity on the integrated assay is that the integrated assay is a standard Seurat assay, rather than a Signac ChromatinAssay, and so is not linked to any scATAC-seq information.

[millersan]

I try the merge vignette, but it seems like 2 sample have very strong batch correction.
That's why I try to used integrated assay. Could you help me ? Thank you so much!