GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2005-02 > 1107406836


From: (John Chandler)
Subject: Re: [DNA] Extinction Rates for new mutations from a Founder
Date: Thu, 3 Feb 2005 00:00:44 -0500 (EST)
References: <002501c5088e$f887e140$eb409145@Ken1> <REME20050201201207@alum.mit.edu> <016901c508e3$f512de20$eb409145@Ken1>
In-Reply-To: <016901c508e3$f512de20$eb409145@Ken1> (knordtvedt@bresnan.net)


Ken wrote:
> If I were to sequentially run 30,000 computer runs of 100 generations

Yes, I admit that's what I really should have done, but it was so
tempting to do all the runs in parallel and save on bookkeeping.

Unfortunately, I overlooked an important factor in the combination
of many runs into one -- my "grand variance" of the whole set was
taken with respect to the mean haplotype of all lines thrown into
a heap, which effectively wiped out all chances of genetic drift.
Observe: if the original founder had one son, and that one son had
a mutation, then the "effective" founding haplotype is that of the
son, and we have instant genetic drift. In such a case, the mutation
doesn't increase the variance of the descendant pool at all.

Ok, so I have to bite the bullet and do the extra bookkeeping. The
result is different, and it appears that 30,000 runs may not be enough
to get a clean indication of what the expected effective mutation rate
is. However, I'll report the result as it stands and try a bigger run
overnight.

Surviving founder lines: 3377 of the original 30,000 (predicted 3429
from the extinction rate).
Mean effective rate of the 3377 cases: 0.0017 (compared to 0.003 actual)
Peak of the distribution of effective rates: 0

Bottom line: I was wrong. There is a reduced effective rate.
However, the indication from this simulation is that the reduction is
not as much as a factor of two. Of course, there is still the other
side of the coin: the reduced effective time span due to bottlenecks.

John Chandler

This thread: