Y-DNA-HAPLOGROUP-I-L Archives

From: "Ken Nordtvedt" <>
Subject: [DNA-HG-I] TMRCAs in Perspective
Date: Tue, 4 Dec 2007 15:54:42 -0700

Tools like TMRCA estimates between haplotype pairs can add information to situations, but the output of such tools can also be overvalued as well. The tools in general use are based on some simplifying assumptions which are not necessarily valid. One particular simplification fed into these TMRCA tools is that the distribution of haplotypes available to be the MRCA is unchanging from generation to generation. We now know from several years of organizing the haplotypes in databases that they come to us as clustered --- as they should be if they descend from clade founders who existed several thousand years ago. The actual distribution of haplotypes in the populations of succeeding generations is in constant flux. The general direction of change for this distribution is the spreading of the haplotypes from their founder's original form --- the clade's modal haplotype --- each generation. So haplotypes further from each clade's founding (modal) haplotype tend to increase in frequency and haplotypes near the modal haplotype of each clade tend to decrease in frequency as time goes by.

One general rule emerges by taking this constant flux in the haplotype distribution into account. If a pair of haplotypes sit close to the modal haplotype of the clade in which the pair exists, then the actual TMRCA generational probability curve is pushed further into the past. On the other hand, if a pair of haplotypes sit far from the modal haplotype of the clade, out on the clade's distributional fringe, then the TMRCA generational probability curve is pushed up toward the present. In the extreme cases these curves can be pushed twice as deep in time or halved in time. Two haplotypes not from the same clade should probably not be (and are not often) used in a TMRCA tool in any event.

Computer software could be produced to offer a more realistic TMRCA estimator so as to take this dynamics of the underlying haplotype distribution into account. It would be an appreciable undertaking, because the data for detailed distribution of the haplotypes would need to be fed into the process. And even then; use of such software would be more complex than using the simplified TMRCA tools. Several months ago such software was produced and tried out for just one robust clade of haplotypes, so these implementation issues became apparent.

So the "One general rule" stated above is perhaps the main caveat to keep in mind when TMRCA estimates are obtained from the standard tools.

Ken

This thread:

• [DNA-HG-I] TMRCAs in Perspective by "Ken Nordtvedt" <>