No announcement yet.

Big Breakthrough in TMRCA calculation

  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Originally posted by benowicz View Post
    . . .. I'm not sure whether I am interpreting your description correctly--that the first record of your earliest known ancestors is 1771? Or that those earliest known ancestors were each born around 1771? Or that you have definitely identified their common ancestor and that he was born in 1771? Any way you look at it, both predictions are pretty close to one another, so that is maybe no big whoop.

    But again, all contingent on understanding the testing resolution of each donor.
    If it turns out that the common ancestor of the I-FT21770 guys was born 1771, it might be interesting to note that my model returns an estimated birth year for their MRCA of 1760 A.D.--assuming everyone, even the I-A11036 guy, is on the BigY-700 platform. Which places 1771 at 45% on the probability curve.

    That case would be one of those where my model performs better than traditional simple mathematical averaging when nested because it specifies the effective mutation rate on the level of the individual sampled item. Which averaging obviously does not. At the 50% confidence level, 1760 A.D. for my model vs. 1799 A.D. for simple averaging for FT21770. I suppose this result should be clear when noticing the really large discrepancy between unshared variants for each clade under A11036. Again, assuming they're all at the same resolution.

    My model prevents the stupid cases you see where the estimated 50% confidence point of a descendant clade is more remote than that of an ancestral clade. This is the benefit of probalistic nesting. A bit more labor intensive, but way more contextually coherent.
    Last edited by benowicz; 22 June 2021, 11:53 AM.