A member wants his son's ....

[dsm]

A member of a group I manage has asked if his son's kit (not yet validated for our project SNP) added into the fully tested sub-group that the father has (the father got his positive results 2 days ago).

He asked for this based on 'by association'. I would like to oblige but commonsense tells me this is unacceptable until his son's results state that he has the required test results. This might be a very touchy matter to some people.

My question is in a broader sense, how flexible we can, or should be, at doing something that in a purely scientific sense, is unproven.

I have seen one other group administrator place (his own nephew) into a tested category that the nephew had not been tested for with the comment 'by asociation'. I am not really at all happy about accepting 'by association' as proof someone has a particular result and is justifiably able to be added as an accepted statistic in that tested clade/category.

Any advice most welcome. Hopefully there is a policy statement that I can refer to.

Thanks

DSM
DF100-CTS4528 project

[efgen]

It's up to each project admin to decide how to group members in the project. Different projects have different goals and different methods of grouping members, so there's no policy for it.

I don't know the specific goals of your project, but unless you plan to publish a study where the data needs to be 100% accurate and you can't use assumptions, then I think perhaps you could bend the rules a little here.

If you were talking about STR mutations, then I would say that you can't assume father and son won't have any mutations between them.

But with SNPs, it's a very reasonable assumption that the son has at least the same SNPs as the father.

So I think it's unnecessary to force the son to spend extra money to confirm that his SNP results are the same as his father's.

Of course, like I said, I don't know the goals of your specific project. But if this was a situation in my haplogroup project, I would definitely be comfortable with making this assumption in the case of a father and son.

Elise

[flanged]

Googling "estimate of NPE per generation" turned up some interesting views with estimates ranging from 1% to 10% per generation, with many settling on about 3%.

Group administrators should take member's claims of 'known' relationships with a grain of salt.
I would go out on a limb to suggest that quite a few of (my surname group's) member's purported lineages are incorrect.

$0.02

[efgen]

flanged,

We're talking about a father and son here. As long as their Y-DNA marker values match, there's no NPE. SNPs aren't going to change that.

Elise

[Wing_Genealogis]

Flip side of the coin

I tend to think about it (as a co-administrator for the R1b-U106 Haplogroup Project) in the opposite way:

To me, it would be a waste of money to have a family member SNP test. The STR results will show whether or not an NPE occurred, and if the STR results are consistent, then they obviously would fall into the same subclade. I would much rather they use this money either to increase the STR markers tested, or donate to the project to give others the ability to increase testing.

A somewhat related, but somewhat tangential issue is that having multiple family members join a haplogroup project has the potential to distort some of the mathematical work which haplogroup projects sometimes work on.

Within R1b-U106, we have a fairly small subclade which is dominated by several members of the same family. When doing calculations (like computing modal and likely ancestral values, variance within the clade and estimating the age of the clade) we need to only include one representative from the family, or else they skew the results.

[1798]

flanged,

We're talking about a father and son here. As long as their Y-DNA marker values match, there's no NPE. SNPs aren't going to change that.

Elise

They must match in SNP not ystr values.

[dsm]

Solution adopted

Thanks everyone for the suggestions & comments.

What I did was to create a new sub-group and named it with the particular SNPs as follows

sub-group 4
CTS4528+ DF100+ Assumed (based on parent STRs(12) matching)

I coloured it green along with the main category
sub-group 3
CTS4528+ DF100+

This as best I can see meets enough of the needs of the member in question and my own wariness at putting someone into a category that I do not have 'adequate' proof of. If the son's haplogroup were listed as R-CTS4528 (like his father) I would have much less hesitation but, as it is the son's haplogroup is listed as R-M269.

Again thanks for the comments & advice.

DSM

[efgen]

They must match in SNP not ystr values.

Sigh, no.

In the past year or so, I've seen more and more people make a blanket statement saying that STR results can't be trusted without SNP results too -- even those at 37 markers and higher.

While that certainly can be the case for some extremely common STR haplotypes and for STR matches that are on or very close to the border of what's considered a match by FTDNA (ie, 33/37, 60/67, 101/111), it most definitely doesn't apply across the board.

It certainly doesn't apply in the case of a father and son, as in dsm's original question. Father and son already knew their relationship and their STR match supports it. Period. Y-DNA SNP testing to prove *paternity* would be futile and a complete waste of money. If they have any need for definitive proof of paternity, then they should take an autosomal test, not purchase more Y-DNA SNPs.

IMO, comments about the need for SNP testing (long before this thread) have gotten out of hand. This has turned into a game of "telephone" where the message keeps getting morphed into something very different from the original intended message. We need to get back to the point where comments like "We're an exact match on 67 markers! Exciting!" aren't immediately met with a retort such as "that's nice, but your match is meaningless unless you spend hundreds of $$ on SNP testing too."

/soapbox

Elise

[efgen]

What I did was to create a new sub-group and named it with the particular SNPs as follows

sub-group 4
CTS4528+ DF100+ Assumed (based on parent STRs(12) matching)

Good compromise

Elise

[lgmayka]

To me, it would be a waste of money to have a family member SNP test. The STR results will show whether or not an NPE occurred, and if the STR results are consistent, then they obviously would fall into the same subclade. I would much rather they use this money either to increase the STR markers tested, or donate to the project to give others the ability to increase testing.

I wholeheartedly agree with this reasoning.

[Jim Barrett]

Sigh, no.

In the past year or so, I've seen more and more people make a blanket statement saying that STR results can't be trusted without SNP results too -- even those at 37 markers and higher.

/soapbox

Elise

Pure fact that needed to be stated!

[lgmayka]

In the past year or so, I've seen more and more people make a blanket statement saying that STR results can't be trusted without SNP results too -- even those at 37 markers and higher.

On the one hand, we have documented cases of 35/37 and 61/67 matches that turned out to belong to very different subclades (U106 vs. P312, L22+ vs. L22-).

On the other hand, such concidences are very rare.

[1798]

Sigh, no.

In the past year or so, I've seen more and more people make a blanket statement saying that STR results can't be trusted without SNP results too -- even those at 37 markers and higher.

While that certainly can be the case for some extremely common STR haplotypes and for STR matches that are on or very close to the border of what's considered a match by FTDNA (ie, 33/37, 60/67, 101/111), it most definitely doesn't apply across the board.

It certainly doesn't apply in the case of a father and son, as in dsm's original question. Father and son already knew their relationship and their STR match supports it. Period. Y-DNA SNP testing to prove *paternity* would be futile and a complete waste of money. If they have any need for definitive proof of paternity, then they should take an autosomal test, not purchase more Y-DNA SNPs.

IMO, comments about the need for SNP testing (long before this thread) have gotten out of hand. This has turned into a game of "telephone" where the message keeps getting morphed into something very different from the original intended message. We need to get back to the point where comments like "We're an exact match on 67 markers! Exciting!" aren't immediately met with a retort such as "that's nice, but your match is meaningless unless you spend hundreds of $$ on SNP testing too."

/soapbox

Elise

I have 77 exact 12 marker matches, 70 of them do not belong in the same subclade. I would have to go back to L11 to the common ancestor. That shows that YSTRs are not reliable.
A father and son should have the same 67 marker hapotype.

[rivergirl]

I have 77 exact 12 marker matches, 70 of them do not belong in the same subclade. I would have to go back to L11 to the common ancestor. That shows that YSTRs are not reliable.
A father and son should have the same 67 marker hapotype.

12 markers is not reliable for men in common haplogroups and subclades.
That does not show that STR markers are not reliable. It shows they need more markers to refine the matches.
How many of your 77 matches are still there at 67 markers?

If all fathers and sons matched exactly there would be no point in using yDNA, as there would never be any mutations. Men everywhere would have the same haplotype.
A father and son must have a mutation at some stage, and it could occur in the present generation as easily as in past generations.

[1798]

12 markers is not reliable for men in common haplogroups and subclades.
That does not show that STR markers are not reliable. It shows they need more markers to refine the matches.
How many of your 77 matches are still there at 67 markers?

If all fathers and sons matched exactly there would be no point in using yDNA, as there would never be any mutations. Men everywhere would have the same haplotype.
A father and son must have a mutation at some stage, and it could occur in the present generation as easily as in past generations.

You are right to a point but the mutation might be in the 200-300 YSTR markers range.I don't think that 67 markers are enough to establish a most recent SNP match.