Closed pinin4fjords closed 4 years ago
yeah, a PR of that kind is welcome. Have you figured out why you need even smaller n_comps
in #76?
Thanks @nh3 , okay, will do. I've been away for the intervening two weeks, so haven't figured out the other thing. But issues with low cell numbers are a bigger issue for us right now than low gene numbers.
With v0.2.9, supplying a matrix with fewer cells than the n_comps setting produces an error. In the following example, input.h5 has 35 cells:
And an error is produced like:
I understand the provision of 50 is non-sensical, but it's a default in the workflow we have, and I'd like that not to break things in cases such as this.
@nh3 - would you be amenable to an analagous PR to https://github.com/ebi-gene-expression-group/scanpy-scripts/pull/76 (which covered a the related issue of low gene numbers), which automatically bumps the n_comps down to the minimum of (default, n_cells, [ n_genes ? ] ).