Is it possible to deconvolute a spatial transcriptomics dataset (10x Visium) into its component cell types using an matched bulk RNA-seq dataset?
Hi @cell_kc, this is not possible via SpatialData, as it is an infrastructure for data storage, manipulation and visualization, but not an analysis framework. What instead the library enables to this regard, is to save the results of the deconvolution into the standard format, and to be able to manipulate it and plot it. Examples of this is given by the “xenium_and_visium” notebooks here: https://github.com/scverse/spatialdata-notebooks/tree/main/notebooks/paper_reproducibility.
The notebooks perform several tasks, but one of these is to integrate the results of deconvolution (using the external tool cell2location) and operate on the joint data together (cropping, aggregation, plotting).