GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-10 > 1159809282


From: "Ken Nordtvedt" <>
Subject: Re: [DNA] MLE
Date: Mon, 2 Oct 2006 11:14:42 -0600
References: <005f01c6e3dd$e0b56660$5102a8c0@PC238587905109><4521382E.5090009@sbcglobal.net>


For those of you for which refering to a natural log table may not be
convenient or transparent, the ln(n/k) factor is just (to sufficient
approximation) "GD divided by n" with GD being number of steps of difference
at considered markers and n is total number of markers considered.

Where the natural log function deviates from the above approximation, I
would not trust the original formula in any case. Remember, this is ONLY
the most likely estimate; without the size of the confidence interval around
it the meaning is not that useful to you.

The mutation rate is the average mutation rate over the considered markers.
Therefore it better depend on which markers are used.

Ken


----- Original Message -----
From: "Alfred A. Aburto Jr." <>
To: <>
Sent: Monday, October 02, 2006 10:02 AM
Subject: Re: [DNA] MLE


> Chris,
>
> > Chris wrote:
>
>>I note several sites ie smgf give MLE's. ie most likely estimate. How is
>>this calculated please (please note I'm no mathematician).
>>
>>Many thanks,
>>Chris
>>
>>
>
> The MLE (Maximum Likelihood Estimate as it is called), in generations,
> is determined by:
>
> MLE = ln(n/k) / (2*mu)
>
> "ln" is the symbol for "natural logorithm"
>
> "n" is the number of markers
>
> "k" is the number of markers whose values match.
>
> "mu" is the mutation rate in generations (generally a nominal value of
> 0.002 is used).
> It looks like the mutation rate used at SMGF depends upon the markers
> used. One
> could work this out though by looking at 1 marker mismatch cases (I
> haven't done it).
>
> On SMGF if you have a 36/37 match (k=36, n=37) then the
> MLE = ln(37/36)/(0.004) ~ 6.8 generations (round to 7).
>
> On SMGF DYS464 is not counted for the TMRCA estimates.
> Al
>
>
> -------------------------------
> To unsubscribe from the list, please send an email to
> with the word 'unsubscribe' without the
> quotes in the subject and the body of the message
>


This thread: