25 thoughts on “Fourteen Percent

  1. This is fun and all, but I have no idea what “14% sub-Saharan African” means. I don’t believe that 14% of the DNA of Cobb (or anyone on the planet) consists of genes found only in people from that region. Can anyone enlighten me?

      Quote  Link

    Report

    • I think you’re right to be a little skeptical. I don’t think that number is exactly what most people think it is (that 14% of his makeup came from some “pure” African contributer of “African” genes). It seems like it’s more of a distance along one axis of a coordinate system that comes from a PCA (or some other dimensionality reduction trick) of a bunch of genetic data.

        Quote  Link

      Report

    • My understanding, and it could very well be flawed, is that mutations amongst the non-coding regions of the genome (the “junk” DNA) tends to be conserved since, being non-coding, it doesn’t affect survival or reproductive success.

      But 14% is an odd number (literally, one seventh) for this sort of thing. To be precisely correct this would require several black ancestors scattered over a few different generations, no?

        Quote  Link

      Report

      • If my understanding is correct, it’s much less concrete than a simple ancestor count. This is hard to describe without getting into the linear algebra, but imagine a coordinate system with two axes. Now imagine we have two genetic markers we’re looking for. We can describe your DNA as being a vector at , , or . Maybe people from one area tend to cluster in one of the four corners while everybody else tends to be in one of the other three. That’s useful for classification. Now imagine a coordinate system with a kajillion axes mapped to a kajillion genetic markers. In that coordinate system, you could plot each person in the world as a vector in that space, and there would be “clusters” everywhere, although a human couldn’t possibly visualize them.

        Most of the axes won’t tell us much (pretty random flat distribution), but some axes will tend to be associated with certain geographical regions, but it’s hard to see because it’s in a kajillion dimensions. There are a lot of ways of dealing with this mathematically, but the ultimate goal is to end up with a new coordinate system with less than a kajillion dimensions, but one that still captures the areas of “interesting” variation. The new axes are (usually linear) combinations of the old axes, which is how you end up with numbers like 14%. We may find that variation along one axis tends to separate people with African ancestry from everybody else, so your distance along that axis is some weighted combination of the presence of genetic markers that were used to create that axis.

        One way to get to the new set of axes would be principal components analysis, but there is a wealth of techniques for separating clusters of data points in high dimensional space. The short version is that, “14% of what?” is a very good question and the answer is likely very subtle–not a percentage of genes or a probability, but a variation along some fairly complex metric.

          Quote  Link

        Report

  2. I expect that if everyone took the same test we’d all be less than 100% whatever we identify as: white, black, red, yellow, etc. I expect you’d even see similiar reactions from Japanese finding out they have some Korean in them or vice versa, etc.

    No, as to his intention of his desire to “turn that town into a neo-Nazi enclave.” I doubt he’ll have any sucess. We had two neo nazis come into my town when I was growing up (out west-more west than ND). They caused quite a bit of gossip. No one went to their store and they left. I expect folks in ND will react the same.

      Quote  Link

    Report

Leave a Reply

Your email address will not be published. Required fields are marked *