In order to generate the reference signature matrix is it ok to use AverageExpression from seurat?

[smk5g5]

Hi,

I have a seurat object which I would like to convert into signature matrix. The data is lognormalized and I am using Seurat's AverageExpression function to generate the signature matrix. I was wondering if this is the correct approach or I should sum the counts for a particular celltype/identity class in the single cell object and then do CPM normalization to generate a better signature matrix for deconvolution?

Best regards!

[ndelrossi7]

Hi @smk5g5! Please refer to this paper for more information on expected input and output for DWLS deconvolution in Giotto.

In addition, we have recently made updates to the suite branch, and you can use makeSignMatrixDWLSfromMatrix() from a matrix that has been pre-processed outside of Giotto. We are in the process of creating a tutorial detailing how to use spatial deconvolution in Giotto that will be out soon!

[skim245]

@ndelrossi7
I'm also trying to use scRNA seq seurat object as a reference to devoncolute cell type to Giotto object, and could not find makeSignMatrixDWLSfromMatrix() function. is this something that will be available in near future? Thank you in advance!

[ndelrossi7]

Hi @skim245, what version of Giotto are you using? That function is available in the Suite branch.