Talk:Haplogroup G2c (Y-DNA)
From Wikipedia, the free encyclopedia
 |
This article is within the scope of WikiProject Human Genetic History, a collaborative effort to improve Wikipedia's coverage of genetic genealogy, genetics-based population history, and associated theory and methods. If you would like to participate, please visit the project page, where you can join the project and/or contribute to the discussion. |
[edit] Standard deviation on tMRCA ?
Can you give a citation for the G5 tMRCA calculation ?
Estimating error-bars for multi-individual tMRCA calculations is a pretty subtle calculation. The methods used in some of the earliest papers - eg the original Cohen Modal Haplotype paper - simply working through the total average squared differences (ASDs) have, I believe, been pretty much completely rejected, for failing to take into account how much of the uncertainty is systematic between different data-points (not random), reflecting the unknown length of the period of shared common ancestry between each pair of data points. AFAIK, any sensible estimates need a pretty full-on wholescale Monte Carlo approach, sampling over different possible tree structures and different possible mutation dates for each step.
Do you know whether that was done in this case?
There is also a wide-open question of the appropriate 'correction factor' (often taken to be about x3) that should be applied to tMRCAs estimated by ASD methods. Disagreement about the size and rationale for this factor is huge. -- Jheald 11:23, 18 September 2007 (UTC)
- Okay, I'm assuming that this network is the basis for the tMRCA claim. There are a number of things wrong here:
- 1. It ignores the very different mutation rates at different markers.
- 2. It appears to use a naive ASD-based calculation to estimate the age of the network.
- 3. The error-bar estimate relies on all the deviations from the mean being independent. They aren't.
- In reality, there is probably room for debate of the actual age of this network over an uncertainty of I would guess about one to three thousand years away from the "best" estimate, if a naive ASD calculation indicates an age of the order of 1000 years. Jheald 15:33, 19 September 2007 (UTC)
