d-callan
commented
1 year ago An alternative might be a Bubble Plot, where we keep the points per sample record, but scale them in size to represent the number of specimens belonging to the sample. This might mean different entities had different floating vizs available to them.
We should probably consider what to do about the outlier points in Box as well...
Example being the sample entity in the megastudy again. The plan is to update floating plots to be able to do math in a way that reflects specimen counts rather than sample or assay counts. If we were to be consistent wed try to do that for scatter too, but that might be difficult or impossible to do well. Theres only so much you can extrapolate about individuals given data about a pool.... Plus even if we did manage it (somehow...) it wouldnt be performant unless heavily filtered.
Do we want to drop scatter? Or only allow it for certain entities? or other ideas?