search

satijalab / sctransform

R package for modeling single cell UMI expression data using regularized negative binomial regression
GNU General Public License v3.0
208 stars 33 forks source link

Repeat on splced counts same normalization as done on counts #112

Closed aghr closed 2 years ago

aghr commented 2 years ago

Dear SCTransform team, Working with Seurat I have three assays: RNA, spliced, unspliced. Applying SCTransform on RNA gives the new assay SCT. Then I would like to apply SCTransform on the assays spliced and unspliced to given SCT.spliced and SCP.unspliced where the same normalization is applied that was previously determined for RNA. I do not want SCTransform to re-estimate the normalization on spliced and unspliced. Ideally it would take the determined parameters from normalizing RNA and applies these to spliced and unspliced, too.

Background: I want the counts, spliced and unspliced treated (normalized) exactly the same to then do RNA velocity and map its results on the UMAM learned from counts. I would like to avoid counts, spliced, and unspliced to be changed (normalized) differently.

Is that possible?

Many thanks Andre

ChristophH commented 2 years ago

Yes, that should be possible. Have a look at the assay, new.assay.name, and reference.SCT.model parameters of the Seurat::SCTransform function. After running SCTransform on the RNA assay, the resulting SCT assay should have a SCTModel.list slot. The first entry would be the reference SCT model that you could re-use. If you have trouble making this work, the Seurat page on GitHub would be a better place to get help.