slowkow
commented
7 months ago To deal with the issue of large matrices, you might consider selecting a random subset of your cells. Or, consider creating a pseudobulk matrix and working with that instead of the single-cell matrix.
Also, I think porting the code to python would be a valuable contribution to the community.
Hi, this is really a practical tool! I was wondering if python users(scanpy users) could benefit from this, particularlly when Seurat can not handle million-scale cells caused by a known issue, i.e., sparse matrices from Matrix package used in Seurat to store counts allow max 2^31-1 non-zero values. Thanks in advance!