Stream: terminology
Topic: de facto compositional grammar
Joel Schneider (Jul 18 2018 at 18:48):
I'm wrestling with a slightly gnarly terminology scenario regarding HLA nomenclature and related code systems, and am interested in feedback on how to best organize the complexity within a HL7 FHIR setting.
The human leukocyte antigen (HLA) system enables the immune system to differentiate self and non-self cells. For medical scenarios involving blood stem cell transplantation, it is essential that the HLA tissue type of the donor matches that of the recipient.
The official IMGT/HLA nomenclature (https://www.ebi.ac.uk/ipd/imgt/hla) is commonly used for transmitting HLA data between institutions.
Nomenclature-level HLA data is often ambiguous. GL String notation (http://dx.doi.org/10.1111%2Ftan.12150) is a grammar which can be used to encode ambiguous HLA data.
NMDP multiple-allele code (MAC) designations, a pre-coordinated code system derived from the IMGT/HLA nomenclature, are also commonly used for encoding ambiguous HLA data.
(IMGT/HLA nomenclature and NMDP MAC designations are the main HLA-related code systems currently in operational use.)
With regard to handling nomenclature-level HLA data within a HL7 FHIR setting, the two main stumbling points I can see are: the lack of an official endorsement of GL String grammar by the relevant code system owner(s), and the need for GL String expressions that combine concepts originating from multiple code systems.
Below is a terse outline that attempts to summarize the problem space.
1. Grammar not (yet?) officially endorsed by code system owner(s).
1.1 FHIR R4 -- use code system supplement to handle this (per recent in-person discussions)
1.2 FHIR STU3 -- extension needed? (annotate CodeSystem? override CodeSystem.compositional?)
1.3 FHIR DSTU2 -- extension needed? (annotate ValueSet? add ValueSet.compositional?)
2. Need for grammatical expressions that combine concepts originating from multiple code systems.
2.1 Relevant code systems
2.1.1 IMGT/HLA -- https://www.ebi.ac.uk/ipd/imgt/hla
2.1.2 NMDP MAC Designation -- https://bioinformatics.bethematchclinical.org/hla-resources/allele-codes/
2.1.3 WMDA additional codes (XXXX, NNNN, UUUU, NEW) -- https://doi.org/10.1038/sj.bmt.1705672
2.1.4 GFE (gene feature enumeration) -- https://doi.org/10.1016/j.humimm.2015.09.016
2.1 Define ValueSet spanning relevant code systems.
2.2 Annotate ValueSet to indicate use of GL String grammar across all its concepts.
2.3 Concept collision -- not an issue yet (for the present set of code systems), but what if it happens?
3. Some real-world GL String usage examples.
3.1 MAC UI, e.g. decode HLA-A*01:AB -- https://hml.nmdp.org/MacUI/
3.2 WMDA Validation simulated cohort (supplemental materials) -- https://doi.org/10.1111/tan.12817
3.3 LOINC HLA-C example answer list -- https://s.details.loinc.org/LOINC/77636-9.html?sections=Comprehensive
3.4 GL Service -- https://gl.nmdp.org
Questions ...
Q1. Is there an established precedent for handling of a de facto grammar not officially endorsed by the code system publisher?
Q2. Is there an established precedent for dealing with a compositional grammar that spans multiple code systems?
Q3. How might things improve if we're able to obtain assent/endorsement from the relevant code system owner(s)?
Lloyd McKenzie (Jul 18 2018 at 18:51):
@ Bob Milius
Joel Schneider (Jul 18 2018 at 20:16):
@Martin Maiers
Michael Lawley (Jul 20 2018 at 01:56):
@Alejandro Metke - has looked at cross-code system issues in the context of OWL (and specifically HPO)
Robert McClure (Jul 23 2018 at 14:41):
@Joel Schneider I've not spent time yet with what you sent so can't give concrete answers, but given we are talking about external code systems, the HTA should be asked to give you help. I suggest planning a presentation of the issues you've noted to HTA. Send a request to discuss to termauth@lists.hl7.org and copy Heather Grain who is trhe chair: heather@lginformatics.com. Calls Thursdays 4p ET q2wk, next one is Aug 2.
Joel Schneider (Jul 24 2018 at 14:06):
Thanks Rob. I've sent an email to the HTA list.
Joel Schneider (Jul 31 2018 at 20:25):
Update -- during a local hackathon last week, with vital assistance from @Lloyd McKenzie, I think we now have a solid plan to move forward on this.
In short, we plan to define a new code system which will incorporate the needed grammar.
Last updated: Apr 12 2022 at 19:14 UTC
