Open cathalgking opened 11 months ago
Hi @cathalgking
You can use cell2location as long as you can provide reference cell type signatures, a matrix of cell types times genes, which contains average RNA abundance in a linear scale. It needs to be normalised for technical effects across bulk samples. Can you derive this from your bulk data?
Normalisation in bulk data is often done in log-space - so normalised RNA abundance values have to be exponentiated before they can be used as input to cell2location or any other spatial data decomposition method.
Also, note that the references for decomposing spatial transcriptomics data need to be quite comprehensive regardless of the spatial data decomposition method. This means that you can easily get artefacts when the reference does not represent full cell type heterogeneity - eg if reference has 3-5 cell types.
I would be also keen to know if this works!
The title is the Q really - I have matched spatial transcriptomics data and bulk RNA-seq data and want to deconvolute the spatial data with the bulk data. Is this possible with cell2location or some other method?