maternity,paternity

[f.a.]

Isn’t possible to use mtdna as a proof for maternity and y- chromosome dna as a proof for paternity . If my mother mtdna doesn’t match mine , is that a proof that she is not my biological mother . If my father y-chromosome dna doesn’t match my bother y-chromosome dna , is that a proof that he is not his biological father .

[JSW]

Re: maternity,paternity

Originally posted by f.a.
Isn’t possible to use mtdna as a proof for maternity and y- chromosome dna as a proof for paternity . If my mother mtdna doesn’t match mine , is that a proof that she is not my biological mother . If my father y-chromosome dna doesn’t match my bother y-chromosome dna , is that a proof that he is not his biological father .

There are two ways to look at it plus the "degree" of proof you
are looking for.
You opening question is on "proof for paternity "
an the answer is one can make stong case but not
"proof" This is because someone else, who
just happened to have the same Y-DNA or mtDNA, might be
the parent.
That is why the real paternity tests use other tests.

Your following examples/questions were on proof of "non-paternity"
That is the other side of he coin and the answer is
YES - more certain for mtDNA than Y-DNA.
Two brothers could have a Y-DNA genetic distance of as much as
five. (one single mutation for one son and one double mutation
for the second son) But if it goes beyound that the odds against
relatedness go way up into the millions to one area.
One can even calculate the odds of the two being related
because we know the statistics.
But if they are really distant it will prove non-paternity.

[whitkeen]

If brothers can have a genetic distance of 5, then it would seem that the 12-marker test is much too inclusive. In our project we have a lot of people separated by 3 and one by 5 from the best researched line. I had assumed we were not closely related to each other. Doesn't that much distance mean not related within a genealogically meaningful time frame? Should I rethink my analysis?

[JSW]

Originally posted by whitkeen
If brothers can have a genetic distance of 5, then it would seem that the 12-marker test is much too inclusive. In our project we have a lot of people separated by 3 and one by 5 from the best researched line. I had assumed we were not closely related to each other. Doesn't that much distance mean not related within a genealogically meaningful time frame? Should I rethink my analysis?

It all has to do with how "sure" you want to be.
The odds of two brothers being different by 2 markers out
of 12 is about 1 in 1,000.
The odds of two brother being different by distance of 5
[one 1 the other a 2] is about one in 60,000.

When people look for "proof" the typical numbers that we
see are a million to one or more.
So 60,000 to 1 would not be "proof"

For a family genealogy of 4th cousins the odds of
a no changes is 78%
of one single step change it is 19%
and a 2% change is 2%

Statisticians like this 95% point.
That may or may not be good enough for a family tree.

Lets take your 3 one step differences on a 12 marker test
Lets assume 5 births to MRCA.

There is one chance in 500 they are related that way.
Only the people who will be spending the money
can decide if it is worth the upgrade to 25 markers to MAYBE
have a better answer

I say MAYBE because they have to beat the odds to get that
better answer.
The most likly out come of the upgrade to 25 markes is
to find 2 more one step mismatches.
Now we are 4 off at 25 and the odds they are related
is still one in 500.
But that is the statistics and until one has the test results in hand
it is all [somewhat wild] speculation.
As they say in the gas milage for cars - your results may very.

If one were to say lets use 1%, that is one in 100, not 95%
And we use 10 generations not 5.
Then the 3 marker mismatch would be "possible" by that standard. [the odds work out to be 88 to 1 against it but
that is better than 100 to 1 and thus possible.]

Bottom line is we each can decide where to draw the line
and in my book that is a good point.

[whitkeen]

Thanks John, I guess. I will try to get a better handle on these odds, and then, once I understand it, I can explain it to the project participants. We know we have several lines that diverged before the early 1600s and came to America or Canada. I was trying to find out if we linked back in England. That's 10-13 generations (for real, not estimated). What are the chances that there will be 5 different lines, off by anywhere from 1 (I consider this one probably closely related) to 5 markers. Incidentally, the two well-researched lines match internally except for the one-off.
It's the others who are not sure if they connect. What are the odds that they do, given a lack of paper evidence to link them with the well-researched ones. But they do have the same surname, English ancestry, and a tradition of being in the same family soo I would not consider them random R1b matches.

[JSW]

re ".... if we linked back in England. That's 10-13 generations (for real, not estimated). What are the chances that there will be 5 different lines, off by anywhere from 1 (I consider this one probably closely related) to 5 markers."

To solve this question you must determine what you think the
haplotype of the MRCA is back in the 1600.
Then for each line your research look at the number of mutation
from MRCA to today.
The haplotype for the MRCA may not be a 12/12 nor a 25/25 match to any of todays men.
Here are the chances for mutations on 12 and 25 markers
13 generations [12 births]; mutation rate 0.002
Num. Num Markers
Mutations 12 25
0 75% 55%
1 21% 33%
2 3% 10%
3 0.2% 2%
4 0.02% 0.2%

RE: "What are the odds that they do, given a lack of paper evidence to link them with the well-researched ones."

Given the above you should be able to pick the odds from the
table I gave.
If you have a different situation I can calculte those odds also

[whitkeen]

Now I know why I did so badly in statistics. Better you than me.
Thanks for the information. I will try to assimilate it and keep it against the possibility that we go for 25 marker tests down the road. Since most of the family come from England, possibly western England, and I think the results they have gotten so far presume R1b, the non-matches don't necessarily prove relatedness. None have actually been tested for haplotype.