Hi,
I have proliferation score for each cell for multiple samples including one control sample. I want to determine what percentage of cells relative to control cells are proliferating in my treatment sample. Is there any suggestions on how to proceed. My thought was to use the Otsu method to find an objective threshold for each sample based on the proliferation score for cells within that sample. This would find a proliferation score threshold that would divide the cells into two groups, proliferating vs non proliferating. Then I can calculate the percentage of cells within that sample that are proliferating. I can do this for each sample including control and compare the percentages of cells proliferating in my treatment samples to the control. Does this sound reasonable or is there a better way to approach this. Appreciate any thoughts.
I would also visualize "the scores", stratified by sample, cell type, basically whatever feature that might be of interest. This can help you to set/review thresholds.
Thank you for your reply. I have the cell types annotated and then for each cell type I have the proliferation scores for each sample. My question is, is the otsu method fine for finding the threshold for each cell type within each sample and then calculating the proportion of cells above that threshold as proliferating. Doing this for each sample to get a percentage of proliferating cells and compare that across samples or is there another statistical way to go about this.
i have a same problem! have you done it? how about the effect? could you share your code with me? thank you very much!!!
I actually did not end up using the otsu method. The scores do not seem to be bimodal so it would be tough. I ended up plotting the proliferation score for each cell and then overlayed seurat's cell cycle calls (G1, G2M, S). When overlayed, I looked at the percentage of cells that were above the highest score that was being called G1. Hope that makes sense. Please reach out if you would like more help