Scaling the dot size based on number of points

It’d be really nice to decrease the dotsize when visualizing large datasets and increase it for smaller ones.

We have a simple scaling in scanpy (https://github.com/theislab/scanpy/blob/master/scanpy/plotting/_anndata.py#L227) which is 120000 / n where n is the number of cells. This leads to very big dots when dataset is very small, so you might want to limit it via min() or max().

[liaprins-czi]

Yeah I think that ^^^ would be good. It’d be confusing to have the deselected size scale up for a small dataset and end up the same size as a default dot of a large dataset.

[colinmegill]

@bkmartinjr @liaprins-czi this will need to take into consideration all three dot sizes:

• default
• deselected
• hover