Implement terms for Population Group value set in NCIt

[mbrush]

As part of our efforts to evolve SEPIO to support ClinGen variant interpretation data, we need classes for 'Population' types described in the gnomAD/Broad, IGSR, and ESP exome sequencing databases.

We would like to use the NCIt as the home for these classes, as the 'Population Group' hierarchy here seems broad enough to cover all population types defined in these systems based on race, ethnicity, and geography. Many terms will map to existing classes, but for many new terms would have to be created in the NCIt.

Lyuba Remennik from the NCIt's EVS has indicated that they are interested in expanding the representation of population groups in NCIt to support epidemiologic research as well as the looming Data Commons. So we can work with the NCIt and other efforts needing Population/Group value sets in this work.

1. African/African American
2. Latino
3. Ashkenazi Jewish
4. East Asian
5. Finnish
6. Non-Finnish European
7. South Asian
8. European-American
9. African-American
10. African Caribbeans in Barbados
11. African
12. Ad Mixed American
13. Americans of African Ancestry in SW USA
14. Bengali from Bangladesh
15. Chinese Dai in Xishuangbanna, China
16. Utah Residents (CEPH) with Northern and Western European Ancestry
17. Han Chinese in Bejing, China
18. Southern Han Chinese
19. Colombians from Medellin, Colombia
20. East Asian
21. Esan in Nigeria
22. European
23. Finnish in Finland
24. British in England and Scotland
25. Gujarati Indian from Houston, Texas
26. Gambian in Western Divisions in the Gambia
27. Iberian Population in Spain
28. Indian Telugu from the UK
29. Japanese in Tokyo, Japan
30. Kinh in Ho Chi Minh City, Vietnam
31. Luhya in Webuye, Kenya
32. Mende in Sierra Leone
33. Mexican Ancestry from Los Angeles USA
34. Peruvians from Lima, Peru
35. Punjabi from Lahore, Pakistan
36. Puerto Ricans from Puerto Rico
37. South Asian
38. Sri Lankan Tamil from the UK
39. Toscani in Italia
40. Yoruba in Ibadan, Nigeria
41. Combined Population
42. Overall Population
43. Other

[mbrush]

Next Steps:

1. Validate ClinGen model use case: confirm with ClinGen that they still need all terms in their Population value set.
2. Identify other partners/use cases: Reach out to other efforts (in ClinGen and beyond) with similar use cases for Population/Group value sets, and collect requirements/terms from them
3. Evaluate NCIt: review current Population Group hierarchy and determine how will it supports these use cases, and what changes would be needed.
4. Improve NCIt: Work with NCIt to extend/refine population group representation.

[larrybabb]

Matt - The reason why ClinGen needs these population types is to allow them to support the ACMG guidelines when applying specific population allele frequency data as evidence for the criterion that requires pop allele frequency data to be met. The ACMG guidelines specifically references ExAC, 1000Genomes and ESP. If we do not have codes for the potentially referenced pop types then we would not be able to represent these 3 repos of pop allele frequency in our variant assessment. So, 1. is not just for ClinGen it is for any group that wants to apply the ACMG guidelines and do so while being able to represent the data in these 3 repos. Is that reasonable enough to cover #1 and #2 above? Or would you need to hear this from an ACMG representative. I have heard that ESP data is not necessarily as useful. I think this is since the emergence of gnomad and other sources of pop allele freq data. But, there may still be those that want to follow the ACMG recommendation by the book, so even if this is so, it makes sense to provide supporting codes for the population types that can be applied computationally for this data.

[mbrush]

Thanks Larry - was mainly recording this ticket for the benefit of Alice Popejoy and Lyuba Remennik as context for my upcoming calls with them., where we will work to get these terms into the NCIt. In the meantime I will put placeholders in SEPIO, and keep you posted when/if we get NCIt replacements.