AnthroScape

Also, it might be better for the comparison to be between the user and averages of each populations since there are often outliers in populations. I don’t know whether that would add too much work or not.

You are right. It would be easier. I'm just a little hesitant with it, because I also really don't want to oversell results. And I think it clearly illustrates the limitations, when your closest match is, say, Dai ethnicity, but then there's two other ethnicities between you and the next Dai. Limitation in any DNA ancestry test, not just this implementation, I mean. I really worry that some of the ancestry calls out there are very exaggerated in their precision claims. I really liked this blog post, where the author also tries to stress test some of the ancestry assumptions. And he does find problems. Anyway, I'll look into it. Maybe it's doable in a neat way.

Most ancestry services out there provide admixture results, which causes a lot of problems because there’s no standard definition of ethnic categories. “Chinese” in one classification system means something entirely different in another for example. The result is totally different admixture on different tests. Inconsistent reference samples, definitions, and classification criteria causes confusion.

But these differences are what each service are promoting to differentiate their products, so there’s no incentive to standardize methods.

Thanks for your work, a lot of nice and fairly accurate stuff for free.

By the way can you explain this? Am I a genius or just smart

jesus wrote:Thanks for your work, a lot of nice and fairly accurate stuff for free.

By the way can you explain this? Am I a genius or just smart

Haha, yeah you must be a genius. My own score is only higher than 82% of pop, so I think actually you should be the one explaining it to me

...but seriously. I'd love to discuss the intelligence terms, because they are the ones I'm working with right now. Your screenshot is educational attainment, from the GWAS module. The Rietveld et al study, from 2014. I left that one in, because - hey, it's just education and it is a reasonably sound study. But I actually have actively been avoiding the intelligence studies untill now for two reasons: 1) that trait really have the potential to offend people - and I get plenty of offended people already, and 2) I actually think there are some underlying problems with the intelligence studies in general.

To explain better - there is actually a 'hidden' module for intelligence and emotional intelligence (IQ and EQ). Two very recent studies, got a lot of attention in the media (Hill et al and Warrier et al). You can check your stats yourself at this impute.me-link. But there's a reason I didn't publish the link on-site already, because I suspect the results are driven so much by ethnicity, rather than, you know, actual intelligence. In other words; I believe impute.me's polygenic score calculator is working correctly for the SNP-lists as published in these two studies. But I think there must be something unaccounted for in the studies themselves, and I can't really understand what. To give an example: all Danish people I test, they get super-high scores. And as much as I'd like to be all-nationalist (I'm Danish), I just don't believe that to be a true reflection of IQ.

So -- long story, but the gist of it is that I honestly don't really believe the intelligence terms enough to release them. At least not without a more solid disclaimer about genetics not explaining everything of a trait, definetly not that one.