search

omnideconv / deconvBench

Comparison of 2nd generation deconvolution methods implemented in omnideconv
2 stars 0 forks source link

Re-define controlled vocabulary #10

Closed kathireinisch closed 9 months ago

kathireinisch commented 2 years ago

Use cell ontology https://www.ebi.ac.uk/ols/ontologies/cl

kathireinisch commented 2 years ago

I created umaps for the singlecell data from lambrechts and maynard with the remapped celltypes (fine level like in the excel sheet) In both datasets the "other cell" doesn't look too god yet, but i'm not sure whether some remapping needs to be done or it's good enough for us. Maybe lambrechts is fine but maynard not? What do you think? umap_singlecell_maynard_otherCells umap_singlecell_maynard umap_singlecell_lambrechts umap_singlecell_lambrechts_otherCells

Here are also the umaps with the "old" celltype annotations umap_singlecell_lambrechts_oldNames umap_singlecell_maynard_oldNames

grst commented 2 years ago

By now, I know that "other" cells in maynard also contain some hepatocytes and glial cells from liver and brain metastases, respectively. I would say for now just pretend the other cells were not there.