CellBender currently depends on Jupyter and relevant libraries (ipython, nbconvert, …). Seeing as only the report functionality depends on Jupyter, it seems a bit odd to make fresh downloads of CellBender take a long time, especially given Jupyter has a quite a large number of dependencies itself.