DNA-R1B1C7-L Archives

Archiver > DNA-R1B1C7 > 2009-04 > 1240410167


From: "Sandy Paterson" <>
Subject: [R-M222] Significance levels
Date: Wed, 22 Apr 2009 15:22:47 +0100
References: <d19.3d757d9f.371fe16f@aol.com> <000001c9c314$34d8b770$9e8a2650$@com><49EECFD4.5010705@aol.com>
In-Reply-To: <49EECFD4.5010705@aol.com>


Some calculations of statistical significance levels :

CDYa = 37

Ewings15/22.6818
Cowans18/25.7200
Other M222+124/380.3263
Total157/427.3677

Is .6818 significantly different to .3263? Using a binomial distribution
with a mean frequency of .3263, the probability of 15/22 or higher is
.000682. So it's significant at the 0.07% level.

Is 18/25 significantly different to .3263? Yes, at the .006% level.

What this suggests is that the probability of Ewings and Cowans exhibiting
15/22 and 18/25 or higher by chance if they were no different from Other
M222+ is less that .07% and .006% respectively.



DYS449 = 31

Ewings21/22.9545
Cowans17/25.6800
Other M222+42/380.1105
Total80/427.1874

Here the suggestion is that the probability of Ewings and Cowans exhibiting
21/22 and 17/25 or higher if they were no different to Other M222+ is
effectively zero.


Both CDYa = 37 AND DYS449 = 31

Ewings and Cowans combined29/47.6170
Other M222+15/380.0395
Total44/427.1030

Once again, using the binomial distribution, it can be shown that the
probability of Ewings and Cowans combined exhibiting 29/47 or more if they
were the same as Other M222+ is effectively zero.

The effects of higher mutation rates are already reflected in the observed
frequencies of the different marker values in the Other M222+ population
(meaning the values .3263, .1105 and .0395. There is no need to worry about
them or allow them to affect conclusions beyond that.

There may be other hypothesis tests worth using. Anyone?


Sandy


















-----Original Message-----
From:
[mailto:] On Behalf Of John Mclaughlin
Sent: 22 April 2009 09:06
To:
Subject: Re: [R-M222] Byrne TMRCA Estimates

<I'd say 295 is a bit more than a handful. Also, I have 22 Ewings and 25
Cowans. They have 708 off-modal matches. You have 20 Ewings and 35
Cowans so you should find far more than the 708 off-modal matches that I
found.

I am counting all matches at a given marker as one match.

As far as I am concerned this:

391 = 10 (2 Cowan 9 Ewing) 2 X 9 18

is one match. It is one mismatching modal M222 marker.

I went through the spreadsheet and added them up line by line. It was a
pain in the butt and something I guarantee I will not do again but I
wanted to see where you were coming from with your 708 off modal matches.

<I'll go through my workings today and produce a table similar to the
above for comparison. At a quick glance though, I see you've left out
DYS491 from the above table although you do refer to it below, where you say

Where is marker 491 to be found? If you're talking about 449 you're
right. I did somehow leave that out of the table. That's the main one
where the two intersect.

Here's a corrected table:

391 = 10 (2 Cowan 9 Ewing)
449 = 31 (22 Cowan, 20 Ewing)
439 = 11 (2 Cowan 1 Ewing)
19 = 15 (1 Cowan, 21 Ewing)
YCAIIb = 22 (2 Cowan, 2 Ewing)
576 = 17 (1 Cowan, 2 Ewing)
570 = 16 (1 Cowan, 1 Ewing)
570 = 18 (2 Cowann, 1 Ewing)
CDYa = 35 (1 Cowan, 3 Ewing)
CDYa = 36 (2 Cowan, 1 Ewing)
CDYa = 37 (16 Cowan, 14 Ewing)
CDYb = 38 (1 Cowan, 17 Ewing)


<You say that other surnames match DYS449 = 31. Sure. Those matches are
included in my figures.

You said you were just comparing Cowans and Ewings.

<The point is that Cowans have far more such matches
with Ewings than the other family groupings. I don't think you can
dismiss that as not being of zero significance.

I don't know if they do or not. I'd have to compare each by hand.
But I don't think this is the point. I think you're looking for a way
to sub divide M222 into sub-clades with this approach and I don't think
it will work. Many of these off modal matches are probably just
parallel mutations and don't mean anything. Especially when one group
only has 2 or 3 out of twenty samples.

< Do you therefore exclude the slower moving markers too?

I'm not excluding any markers except possibly CDYab. But I put that in
the tables too so nothing is excluded.

<Carry on with this
reasoning and you'll logically end up comparing nothing with nothing and
conclude that all conclusions are inconclusive.

Since this is not my line of reasoning I have no concerns about possible
conclusions. Thanks for the warning though.



John












R1b1c7 Research and Links:

http://clanmaclochlainn.com/R1b1c7/
-------------------------------
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: