GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2011-11 > 1320610271


From: "Kenneth Nordtvedt" <>
Subject: Re: [DNA] Any feedback on the RCC method of TMRCA estimates?
Date: Sun, 6 Nov 2011 13:11:11 -0700
References: <20111106134634.FNSYS.1829550.imail@fed1rmwml4201><AF5770CAD8F149EC9EE451556871567C@kenPC><000301cc9cbd$b6db7750$249265f0$@com>
In-Reply-To: <000301cc9cbd$b6db7750$249265f0$@com>


I can use CORREL or create it myself, but I wanted to confirm that's what is
being used. Physical plausibility is sought. Subtracting the mean of the
STR repeat values from each repeat values has no physical sense.

The correl of haplotypes 20,20,21 and 25,25,27 is 1 because of those
subtractions of mean. So 1 minus that correlation is zero!

The correl of haplotypes 20,20,21 and 20,20,22 is 1 because of those
subtractions, etc.

-----Original Message-----
From: Sandy Paterson
Sent: Sunday, November 06, 2011 12:53 PM
To:
Subject: Re: [DNA] Any feedback on the RCC method of TMRCA estimates?

The formula is

r = [sum of (X - mean of X)(Y - mean of Y)] / [SQRT [{sum of (X - mean of
X)^2}{sum of (Y - mean of y)^2}]]

In the old days before electronic calculators, it was practice to use a bit
of algebra to get rid of the means. This had manual computational
advantages.

The formula has plausibility in comparing one variable (say height) with
another variable (say weight).

I'm not sure whether this helps, but if you like, I can set up an Excel
spreadsheet that calculates a correlation coefficient from first principles,
as well as using COREL in Excel, in order to demonstrate that both produce
the same answer. I could also easily show how the paper uses (abuses?) the
concept in trying to apply it to two haplotypes.


Sandy




-----Original Message-----
From:
[mailto:] On Behalf Of Kenneth Nordtvedt
Sent: 06 November 2011 19:02
To:
Subject: Re: [DNA] Any feedback on the RCC method of TMRCA estimates?

I just want the formula. Why would anyone use a method if they could not
write down the formula.
I'll read the paper if the formula has any plausibility.


> Take two 37-marker haplotypes.
> Calculate the correlation coefficient using excel, but pretend that
> instead
> of 2 people with 37 attributes, you have 37 people with 2 attributes.
> Take the reciprocal of the 'correlation coeficient' so calculated.
> Subtract 1.
> Multiply by 10,000.
>
> That's the RCC.

[[ Two haplotypes are h1[i] and h2[i] with i running from 1 to 37. Would
you please convert your comment into the formula for RCC? Excel's correl
function involves some average quantities as well as component values. KN]]




-------------------------------
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


-------------------------------
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: