Hi there, I have run into an issue in dataset integration.
I used celltypist.harmonize to merge the datasets which were pre-annotated under low-hierarchy. the reannotation result turned to have around 1/3 cells annotated as "UNASSIGNED". Then I tired to annotate each dataset under high-hierarchy, which is like 5-6 basic cell-types per dataset, the "UNASSIGNED“ cells decreased, but still take about 20% of total cells. All these "UNASSIGNED" cells were assigned into certain groups. I wonder whether this is normal, or is there any solutions to further decrease the number of "UNASSIGNED" cells?
BTW, thanks for such wonderful tool!
Hi there, I have run into an issue in dataset integration. I used celltypist.harmonize to merge the datasets which were pre-annotated under low-hierarchy. the reannotation result turned to have around 1/3 cells annotated as "UNASSIGNED". Then I tired to annotate each dataset under high-hierarchy, which is like 5-6 basic cell-types per dataset, the "UNASSIGNED“ cells decreased, but still take about 20% of total cells. All these "UNASSIGNED" cells were assigned into certain groups. I wonder whether this is normal, or is there any solutions to further decrease the number of "UNASSIGNED" cells? BTW, thanks for such wonderful tool!