Archiver > GENEALOGY-DNA > 2009-07 > 1248204434

From: Jonathan Day <>
Subject: Re: [DNA] Who killed the men of England -
Date: Tue, 21 Jul 2009 12:27:14 -0700 (PDT)
In-Reply-To: <B52C856797054E98BDB010DACF5498CE@john>

--- On Mon, 7/20/09, Alister John Marsh <> wrote:

> From: Alister John Marsh <>
> Subject: Re: [DNA] Who killed the men of England -
> To:
> Date: Monday, July 20, 2009, 4:25 PM
> Jonathan,
> You said
> >>>>>>>
> I would recommend that Genealogical DNA experts should put
> forward a
> protocol on how many STRs/SNPs need to be tested before it
> is academically
> reasonable to claim such-and-such a person's ancestors were
> in such-and-such
> a migratory wave (rather than simply coming over much
> later). I imagine this
> to be in a table showing a suspected migration Y years ago
> vs. M minimum
> number of markers.
> Too few markers cannot possibly distinguish between a
> short-term flood and a
> long-term trickle.
> Thoughts of crocodiles and archaeologists being dangled
> over them are
> definitely appealing but arguably off-topic.
> <<<<<<<
> I am interested to know which haplogroups and subclades
> were part of various
> archeologically indicated or historical events, but it is
> another thing to
> know simply from a person's haplotype/ haplogroup/ subclade
> if a particular
> persons ancestor was part of a particular migratory event.

You can't "simply know", it's always going to be a balance of probabilities. That's a given. The question is, how much information do you need before you can assign a meaningful probability? Until there is some sort of minimum standard that is set out, we will forever get claims and counter-claims as to what the DNA actually shows. These will inevitably get in the way of any actual science getting done.

> If a haplogroup subclade can be identified as having arisen
> within a
> particular migratory group, it would be possible to say
> that ancestors of
> all of that subclade and downstream sub subclades were part
> of that
> migratory group.

Yes, that is true. A slightly weaker case can be made if you find a haplogroup that can be identified as -likely- to have arisen in the group the migratory group split from AFTER the migration. It's weaker in that it may be a group that migrated back again.

> Proving that a subclade SNP mutation took place in a
> particular migratory
> group is not easy, currently near impossible.  But I
> would not discount the
> possibility that we may over time link some SNPs or even
> some distinctive
> haplotypes with migratory groups in the past 2,000
> years.  In the past 2,000
> years, DNA is not the only tool we have, but we have some
> help from the
> early semi mythical pedigrees stretching back in a few
> cases to near 2,000
> years.

I'm not convinced it's near-impossible. I am convinced that it's not easy, and I'm definitely convinced that if projects like Genographics tested for more than 12 markers, we might have a better understanding of how "not easy" it is. I'm not sure how many markers they'd need to test, or which ones, but if it's possible to reach every possible conclusion and justify it with the available data, 12 is not enough.

> One project I am working on is to try and establish if a
> particular Y-DNA
> cluster descend from a named historic figure living about
> 1400 years ago,
> who in turn is speculated by some to have descended from
> participants in a
> migratory event 1800 years ago.  I may have a low
> chance of success, but at
> least I do have a theoretical chance of success.  If I

I'm not sure why you say you've a low probability of success. You presumably have geographic data as well as genetic data, and maybe some genealogical data as well, and if there's archaeological DNA from the earlier migration, so much the better.

After that, it should be a number-crunching exercise, provided you've a sufficiently large amount of data.

> should succeed, it
> would be possible to identify some people by SNP or
> distinctive haplotype
> with a reasonably probability of descending from a named
> historic figure
> living 1400 years ago, whose ancestors were believed to
> have participated in
> a migratory event about 1800 years ago.

If you succeed, then yes, you will have shown it possible to do what you have done. (In fact, it is a truism that any accomplishment is an existence proof of a solution to a problem. :)

> Archaeology may contribute to proving or corroborating
> where haplogroup
> subclades arose, particularly if we can get to the point of
> reliable DNA
> testing of archaeological specimens,... which seems to be
> coming. 

That is certainly true, with hair samples proving to be better than bone or teeth.

> I would resist feeding archaeologists to crocodiles, as
> "they are a
> necessary part of a team effort".  If you dispose of
> the archaeologists,
> where are we going to get the ancient DNA specimens to test
> by the DNA
> science part of the team? 

Well, yes, but can we at least feed the particularly dense archaeologists to the crocodiles? There's a certain minimum standard in the field, below which results are unlikely to be useful and where potentially valuable information is likely to have been lost.

> Perhaps we could get a mathematician to predict the
> probability of what a
> DNA scientist might find "if" he had some dated
> archaeological specimens to
> test.  That way we could dispense with both
> archaeologists and DNA
> scientists at the same time, and just run with the
> mathematicians
> predictions of probabilities.  Would you feel happier
> with that?  Ken, can
> you run up a quick formula to do that?  Or would we be
> statistically better
> off to keep the archaeologists and DNA scientists in the
> team.

I'd actually prefer it if both geneticists and archaeologists were better-funded and encouraged to use the most powerful techniques available in their disciplines. As it stands, cutting-edge tools rarely get used. In some cases, they're too expensive. In others, unfamiliarity with what is available is a killer.

(I had a fascinating, albeit brief, discussion with the archaeologists who dug at Stonehenge a while back. They were finding lots of material for which dating techniques existed, but they were only going to use C14 dating. This isn't the most accurate method they could have used - dating the pottery by thermoluminescence might well have produced better results - but it was the method they were comfortable with.)

Mathematicians and computer scientists can then produce models for filling in the gaps, providing a "best fit" solution (or more likely a range of solutions) to what should be found where, where data is currently unavailable. This can be tuned and corrected as needed when new discoveries are made. But since it's obviously impossible to obtain either archaeological OR genetic data for each and every year since humanity began, there will always be gaps in both records. Since there's no guarantee the gaps will coincide, you will always need some means of deriving probable values for one based on actual values for the other.

This thread: