0

From a February 18 23andMe tweet:

Only about 2.4% of participants included in large-scale genetic studies have African ancestry.

Is this claim true?

Oddthinking
  • 140,378
  • 46
  • 548
  • 638
  • 1
    *Large scale genetic studies* and the workings of 23andMe (and similar companies) are not the same. There's plenty of information around showing how the databases assembled by these companies are biased towards white caucasian populations (example: https://www.bbc.co.uk/programmes/w3cswh1d) –  Apr 10 '19 at 07:34
  • 2
    Please add a link to the tweet. – Oddthinking Apr 10 '19 at 10:07
  • 7
    I assume this means direct line to modern Africans, and not via Europe, Asia, Oceania, etc. Otherwise it should be 100%, surely. – Oddthinking Apr 10 '19 at 10:09
  • 1
    A link would be https://www.facebook.com/23andMe/photos/only-about-24-of-participants-included-in-large-scale-genetic-studies-have-afric/10158177587577802/ giving the source as https://genomebiology.biomedcentral.com/articles/10.1186/s13059-018-1396-2 Please [edit] this info in. – LangLаngС Apr 10 '19 at 10:30
  • So what if the claim is true? It merely demonstrates that the "studies" did not evenly sample the population of the US. – Daniel R Hicks Apr 10 '19 at 17:19

1 Answers1

2

Yes, this claim is true.

From "A standardized framework for representation of ancestry data in genomics studies, with application to the NHGRI-EBI GWAS Catalog",

African ancestries comprise 2.4 % of individuals but contribute 7 % of associations.

  • @Fizz If you have a better answer, I will accept it. –  Apr 10 '19 at 05:05
  • And if you think that's bad "From the 2011 data, African Americans represented the largest non-European population included in GWAS." https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4418477/ – Fizz Apr 10 '19 at 05:06
  • 1
    @Fizz What are non-European populations? –  Apr 10 '19 at 05:08
  • From the same paper you cited "Similarly, only 1.3 % of individuals in the Catalog are Hispanic or Latin American." – Fizz Apr 10 '19 at 05:17
  • @Fizz From yours, its <1% for both combined. Why the discrepancy? –  Apr 10 '19 at 05:19
  • 1
    It depends on the criteria for including the study in the survey. And it's actually " For ongoing studies (e.g. 2011) in [RePORT](https://report.nih.gov/), GWAS accounted for only ~14%, ~3%, and < 1% for African American, Hispanic, or Jewish ancestry populations, respectively. " The number of studies on African-Americans is actually respectable, but the sample sizes are small. As for the other (US) minorities... not even that can be said. RePORT is based on US NIH grants, which explains their focus. – Fizz Apr 10 '19 at 05:25
  • 1
    But since US research tends to dominate in this area... "Subsequent analysis of published GWAS from PubMed using the same search terms produced 4,942 publications, of which African Americans accounted for ~3% with Hispanics and Jewish accounting for < 1%," – Fizz Apr 10 '19 at 05:27
  • 2
    Also, since quoted a tweet of 23andMe, *commercial* databases are also known for their extreme Caucasian bias: https://qz.com/765879/23andme-has-a-race-problem-when-it-comes-to-ancestry-reports-for-non-whites/ – Fizz Apr 10 '19 at 05:29
  • Currently this only cites the source for the tweet, reiterating the claim. – LangLаngС Apr 10 '19 at 10:29
  • this article also has the 2.4% https://genomebiology.biomedcentral.com/articles/10.1186/s13059-018-1396-2#Fig2 – bukwyrm Apr 11 '19 at 04:59