With the news of the October update on Ancestry, we naturally rushed to check the new estimates. While the update seems to represent progress in a few areas, it's still a bit of a mess with the additional regions that overlap and are sometimes prone to misleading or potentially inaccurate results. We compared our results with those of our Haitian parent to take a closer look at African ancestry in particular.
The new Igboland region is vast and encompasses a large part of Nigeria. Indeed, parts of it on the map extend into the central regions of the country. Nonetheless, we have found Igbo or Igbo-related ancestry to be very likely in our case. The preponderance of "Ibo" captives in Bainet during the 18th century as well as some of the shared genetic matches between us and our Haitian parent with Igbo people in Nigeria seem to confirm it. Yet the Igboland region also includes Igbo, Esan, Idoma, Igala, and Ibibio peoples. If one accepts oral traditions of the Idoma, for instance, then some of these peoples are undoubtedly related to groups further north or had migrated from central Nigeria in centuries past. The Sierra Leone region reference panel is derived solely from Mende, a group we have found shared matches with as well as 23andMe genetic group matches.
The partition of Benin and Togo and the Yorubaland regions are also difficult to interpret. We undoubtedly do have ancestry from parts of that vast region. Our shared genetic matches, however, were with Ga or seemingly Ga peoples. As for some of the new regions like Mali-Dogon Peoples or Mali: Mande Peoples, we cannot help but wonder if the trace scores are more likely de to the genetic overlap of some of these groups with others. Something similar appears to have occurred with the Hausa and Fulani category. Although in past updates the Hausa were part of a "North-Central Nigeria" region which encompassed the Afizere, Anaguta, Atyap, Berom, Ham, and Hausa, the 2026 Update has split that region into new regions. Instead, the Nigeria-Jos Plateau region now includes several of those groups: Berom, Jarawa, Hausa, and Anaguta (plus the Saya and Polci). But by matching the Hausa with Fula, particularly with samples drawn from further west in West Africa, the algorithm seems to be inflating the "Hausa and Fulani Peoples" estimate in many users. Presumably, Fula samples with admixture from Senegambian, Mali, Guinea, or other western countries are being used here and inflating the scores based on shared Senegambian ancestry with various other groups. Perhaps a more likely indication of northern Nigerian ancestry is the Jos Plateau region and Southern Kaduna & Central Plateau region. These regions include groups that were once part of the "North-central Nigeria" region that included the Hausa. As previously stated, the Hausa are also included as one of the groups in the Jos Plateau ethnic groups, too.
Overall, the update was a bit of a confusing mess and some attempts at greater detail or granularity likely muddled things. There are some clear errors but maybe subsequent updates can attempt to improve it. When comparing this with the genetic groups on 23andMe, the latter appears to have the edge in terms of general accuracy or reliability (even if they did lower the matching threshold to increase the number of genetic groups to assign). Ancestry still hasn't taken the next step of indicating which groups in their reference panels users share DNA with, so that can be one area to improve results. it would also be nice if some improvement can be done in these confusing maps which, perhaps inevitably, overlap in confusing ways. And getting data from the Lake Chad region, northern Cameroon, Congo and DRC wouldn't hurt, too. The Luba peoples region is interesting, yet we are lacing other interior groups between them and the ooast.
No comments:
Post a Comment