search

LieberInstitute / spatialDLPFC

spatialDLPFC project involving Visium (n = 30), Visium SPG (n = 4) and snRNA-seq (n = 19) samples
http://research.libd.org/spatialDLPFC/
16 stars 3 forks source link

Deconvolution: Visium IF data #128

Closed lcolladotor closed 1 year ago

lcolladotor commented 2 years ago

Once we have selected the deconvolution marker genes from #127, at broad cell type and layer level resolutions run:

So in the end we'll have 6 sets of results. This will feed into #99.

lcolladotor commented 2 years ago

Based on https://github.com/LieberInstitute/spatialDLPFC/issues/129#issue-1356351532 at #129, can you remind me what is the range of the cellpose-derived number of cells per spot on the Visium IF samples? Maybe with a boxplot by sample. Thanks! I don't remember if any spot exceeded 20 cells per spot. If they do, we might need to check them spatially with vis_grid_gene() http://research.libd.org/spatialLIBD/reference/vis_grid_gene.html similar to what @heenadivecha did at #91 (to check that they come mostly from wrinkles / damaged parts of the tissue) and consider using a cap on the number of cells or dropping spots like discussed at #129.