Benchmarks
We find that the FastProjectR Analysis performs well on input sizes where clustering is not required (i.e. <=35000 cells), where the analysis time scales almost linearly with the number of cells.
We find that the FastProjectR pipeline works extremely well for larger inputs as well; with the clustering algorithm employed, the complexity seems to scale much more linearly with the number of cells. Assuming memory is not a problem storing 1M cells, computational demands shouldn't be too strenuous.
