GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-01 > 1137252549

From: "Alfred A. Aburto Jr." <>
Subject: Re: [DNA] For the more mathematically-inclined on the list - questionon Mode / MLE and Walsh's paper
Date: Sat, 14 Jan 2006 07:29:09 -0800
References: <20060114000151.24499.qmail@web404.biz.mail.mud.yahoo.com>

> Gary Lee wrote:

>In Bruce Walsh's paper, Estimating the time to the MRCA, he presents various calculations for computing MRCA.
>
> In table 1, he shows for a n=50 marker test, a MLE (mode) of 5.1, 10.2 and 15.5 for a 1, 2 and 3 marker mismatch. I can replicate all of his calculations in this table for mean, SD, Median and confidence intervals, but am struggling with how he derives the MLE.
>
> Any ideas from the list?
>
> Thanks in advance.
> Gary Lee
> an-admitted non-Math-major in my undergraduate days
> mathematically
>
>

Ok, I found Walsh's paper :-)

The MLE is given by equation 5:

MLE = (1/(2*mu))*ln(n/k), where
n is the number of markers, for two individuals that match at k out of
n markers, and "mu" is the mutation rate. In the paper Walsh assumes a
mutation rate of mu = 1/500 ...
e.g.: for n=20,k=15, mu=1/500, MLE = 71.9205 generations ...
-----------------------------------------------------------------

I have my own question though (requires a new topic if anyone answers):
What bothers me, being new to this field, is that the MLE is based upon
markers while I think it should be based upon the number of repeats