GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2002-07 > 1025811904


From: "John A Hansen" <>
Subject: RE: [DNA] MRCA calculations, sperm, and phone calls
Date: Thu, 4 Jul 2002 12:45:04 -0700
In-Reply-To: <JCHBN.020703.224433.RC0@CUVMB.CC.COLUMBIA.EDU>


Dear John:

The question about Poisson vs other mathamatical
algoritms is an interesting one. Although the time length
of one generation does affect the mutation rate
per se, the mutation rate is still some function of the number
of the number of actual conceptions ( minus the non paternity
incidents). The related question is the calibration of
that mutation rate with the actual documented genealogy
correlation. So while we have some random generation
math function ( Poisson etc) , there is a calibration factor
that needs to be implemented as well.
So that leads me to suggest that some function of
non-linear differential equation ( allows for step functions
and correction) will be the final modeling solution.

Best Regards
John A Hansen


> -----Original Message-----
> From: John F. Chandler [mailto:]
> Sent: Wednesday, July 03, 2002 7:44 PM
> To:
> Subject: Re: [DNA] MRCA calculations, sperm, and phone calls
>
>
> Gregg wrote:
> > I still don't see any reason that
> > Poisson would not be applicable in the DYS mutation cases. It
> seems to me
> > that you only need to satisfy the Poisson assumptions.
>
> True, and that's exactly the problem here. I'm arguing that germ-line
> mutations do NOT satisfy the assumptions. In particular, Poisson
> assumes that the event in question is equally likely to occur in any
> time span of equal length. The Y DNA mutation transmissions occur only
> at the time of fertilization, i.e., only one opportunity per generation.
> Poisson does not model that well.
>
> > error introduced seems to me to be less than that in the
> uncertainty of the
> > mutation rate itself.
>
> That's exactly what I said the first time around. It always comes back
> to this in the end. Someday, though, we will have enough DNA studies
> tied to deep-rooted pedigrees to really gauge the mutation rate...
>
> John Chandler
>
>
>

This thread: