Get sample ID (not donor ID)

[stemangiola]

Description

I was wondering if there is information in the metadata or elsewhere about sample identifier. Donor identifier is good, but the same donor can have samples for multiple tissues and or disease states.

Thanks a lot.

[ebezzi]

Hi @stemangiola,

we considered the idea but we determined it wasn't feasible with our current schema, since it would need to be standardized across all of our datasets. You can see this issue for more details.

[stemangiola]

Thanks Emanuele,

as we would like to provide an annotation layer to census, where, among other things we heuristically derived a sample label, we would then need the correspondence between the original cell ID and the census cell ID, to transfer the label.

Do you publish it somewhere? Besides this scope, I think that linking would be very important for the community.

Thanks a lot.

[ebezzi]

There is an observation_joinid column that you can use to correlate cells from the Census to cells from the original h5ad in CELLxGENE Discover.

Hope that helps!

[stemangiola]

@myushen