Manually logging TPM count matrix in Seurat

Question

[ww22runner]

Hello everyone,

I have some TPM data for single cell RNA sequencing but I am unsure of how to manually log transform the count matrix in the SeuratObject created. So far, this is what I have:

df <- read.table(file="...",nrows=2000, sep = "\t", row.names=1, header=T)

proj <- CreateSeuratObject(counts = df, project = "proj", min.cells = 3, min.features = 200) 

proj <- subset(proj, subset = nFeature_RNA > 200 & nFeature_RNA < 2500 )

GetAssayData(object = proj) = log(GetAssayData(object = proj)@x)

I am trying to log the tpm counts in the last step but I get an error so I am assuming that I am accessing the data incorrectly. Any help would be greatly appreciated!

Answer 1

[rpolicastro]

This will work.

proj[["RNA"]] <- CreateAssayObject(counts = log(proj[["RNA"]]@counts))

You might want to log+1 transform though, since there are going to be a ton of 0 values that will evaluate to -Infinity in R.

As a suggestion, if you can get a hold of the raw counts, you should use Seurat's SCTransform normalization, which is their preferred normalization scheme. Relevant paper, and the accompanying vignette.

[ww22runner]

Thank you for your reply, when I try this, I get this error:

Error in (function (cl, name, valueClass)  : 
  assignment of an object of class “dgeMatrix” is not valid for @‘counts’ in an object of class “Assay”; is(value, "AnyMatrix") is not TRUE

[rpolicastro]

I edited the answer to fix the error. I confirmed it is now working.

[ww22runner]

Thank you, that worked! Is there any resource that can better help me understand how to access the different components of the Seurat object?

[rpolicastro]

When they updated from version 2 to 3, they created a cheat sheet that is a nice overview of how to access different parts of the data.

[ww22runner]

Yes, I was referring to this: https://satijalab.org/seurat/essential_commands.html from which I gathered that I could access the count data by using GetAssayData(object = object) but I was wondering how I could have derived at CreateAssayObject(counts = ...) or if there is some other documentation I might have missed.

[rpolicastro]

If you type help(package = Seurat) and scroll down a bit, it lists each function in the package with a short description of it. You can then open up the help documentation for any of those functions for more information.

[ww22runner]

Thank you, appreciate it!