Add support for recording annotation type

cellannotation / cell-annotation-schema

General, open-standard schema for cell annotations

9 stars 1 forks source link

Add support for recording annotation type #29

Open dosumis opened 8 months ago

dosumis commented 8 months ago

Some author categories record cell state rather than cell type. Others record a mix of the two, or some other annotation relating to cell properties. It would be useful to be able to systematically record this - maybe at the level of labelsets.

dosumis commented 8 months ago

Once this part of the standard is agreed, we should align the behaviour of pandasaurus_cxg with it.

evanbiederstedt commented 8 months ago

Initial thoughts:

I recognize the value of this. Maybe simply asking users this question would be useful...
This is such a massive fight, I really don't want to get into it....
At some point, I think people will recognize there isn't a major difference between "a cell type" and a "cell state" in transcriptomic space.

dosumis commented 8 months ago

I would like to support this in the standard whether or not CAP chooses to adopt it. It will remain optional. There are clear cases of distinct state and type annotations in CxG and (e.g. whether cells are dividing or not) I would like to be able to distinguish them.

dosumis commented 8 months ago

At some point, I think people will recognize there isn't a major difference between "a cell type" and a "cell state" in transcriptomic space.

I disagree (although to be clear we'd both need to define our terms carefully) - but that is a much bigger discussion.