My geographic ancestry

The interpretation of my mitochondrial DNA given by 23andMe seems misleading. My autosomal DNA is 100% European. My mitochondrial DNA (the female line) is reported 98% American Indian(!). Implied prehistoric global human migrations are problematic. Measurement error is unlikely. More likely is data processing oversimplification.

Here's how it can make sense: Perhaps the correct interpretation of the map should be "this is the probability that these people (here on the map) belong to my group", not "this is the probability that I belong to their group". Big difference! But the more I think about it, the more confused I get!

Some fundamentals

Let us count the number of base pairs (BP).

The chip set used for measurements records one million BP.
Human (autosome) DNA is about 3 billion BP.
Human Y chromosome DNA (male line only) is 60 million BP.
Human mitochondrial DNA (female line only) is 16,595 BP.
The number of "female types" is something like 50. Mathematically the number of female types could be 16,595, but in practice many types do not occur, or occur very rarely so are ignored. My type has the very simple name "X2" which already makes me suspicious they are not modeling enough types. From the published literature I've seen about 50 interpreted types (of the mathematically possible 16,595 types).
By contrast, my male line Y chromosome type is named "E1b1b1a2*". The extra length of the name suggests that the male line is decomposed into many more types, justifiable by the larger size of its genome. My Y map focuses sharply on Morocco, which fits with a silly story I once invented of my ancestry to explain my dark skin and blue eyes: An old Claerbout was a Moor or Jew living in Spain when along came the Spanish Inquisition. He should have gone south but got mixed up and went north instead with the Hollanders. Another fanciful story about my name Claerbout, claer=clear, bout=boutille(French)=bottle. My ancestor figured out how to make the first clear glass bottles, the bottoms of which were to become the microscope of van Leeuwenhoek and the telescope of Galileo. (wink, wink)
Solution to the maternal line problem: X2 splits into X2a (North America) and X2b Morocco! Really! It seems 23andMe simply had not the resolving power.

The word "haplotype" means the one million measurements out of 3 billion possible measurements are supposedly chosen at "good locations". This a large loss of information, I can't guess how much. I don't know how many of the one million measured BP are allocated to mitochondria but it seems like they needed a few more.

Who am I to know anything like this?

I am not a biologist. I invent and test mathematics for reflection seismology to make images of the interior of the earth purchased by oil companies. In my field, as with genomes, the data is voluminous and excellent. It is repeatable to a much greater precision than our ability to formulate mathematical models. A great many scientists are far less fortunate than us.

How is this DNA testing going to be worthwhile for you?

I did find out my probability of prostrate cancer (worse) and diabetes (better) than normal, but that wasn't awfully significant to me. I went into this thinking it might be worthwhile somehow, but I had no idea how. After you sign up, pay, and spit, they involve you with a long, long, list of questions about your body. The more you answer, the more it helps them, and may eventually help you. Finishing this at 1:00am, the last question was,

"Can you think of any other question we should have asked."

My reply was,

"Yes. You should have asked if I had lost a child to sudden cardiac arrest."

I advocated that question because I did lose a wonderful child believed to be perfectly healthy. I have since learned that some cases are genetically predictable and preventable. If they adopt my question, that by itself has paid me generously for joining their data base. It may save someone's child.