search

airr-community / airr-standards

AIRR Community Data Standards
https://docs.airr-community.org
Creative Commons Attribution 4.0 International
35 stars 23 forks source link

Representation of bone-marrow chimerism #334

Open bussec opened 4 years ago

bussec commented 4 years ago

Situation: Allogenic (but not xenogenic) bone-marrow chimeras are not unusual, both in research as well as in clinical settings (e.g., BM transplant as treatment for hematologic disorders).

Problem: In a chimeric sample (e.g., splenocytes or PBMCs) there are cells from two subjects (donor and host), but the AIRR schema assumes an 1-to-N relationship of Subject to Sample. The cells can usually be distinguished by phenotypic markers, but this will only happen in the during cell processing.

Question: Should we allow multiple Subjects per Sample? Or do we explictly consider this case to be out-of-scope?

Note: This issue focuses on a different aspect of BM-chimeric subjects than #137, which deals with annotating the species of the donor. However, the xenograft part of #137 could also be solved here, as it is the more generic approach to chimeras.

schristley commented 4 years ago

@bussec Does such a study exist with publicly available AIRR-seq data? I would consider it worthwhile to curate that study so we have a concrete example.

bcorrie commented 9 months ago

@bussec part of AIRR v2.0? Seems a subtle enough issue that perhaps we can consider it out of scope while at the same time if we implemented it I think it would substantially break the hierarchy that we currently have.

bussec commented 9 months ago

@bcorrie Looking at this again after 4 years (when we were still young and full of dreams), I still think that this is an relevant - although not a super-important - use case. However, I should also note that my younger self was obviously not aware of the fact that we did not and do not have any explicit way of referencing between the Subject and the Sample object (no IDs, no references) except by including instances of these classes into the same Repertoire object. Assuming that we are not going to change this for v2, we would be another way to provide the information about a host's chimerism. This could, e.g., also be provided as part of the Subject record itself, although this would have the disadvantage that we would loose the linkage to the source on the cell level.