DNA RESOURCEs: Romanov-related scientific papers

[Annie]

Dave, please don't even joke about that!     Sorry, I couldn't resist  

Actually it's the 'evil' twin that people don't like, the real Bush is a wonderful guy!

[Annie]

Thanks Annie. That's great.
Actually this DNAi site has lots of other interesting stuff. I never seen such a great computer animation to explain the PCR.

Oh thanks I'll have to dig around and check it out!

[helenazar]

I know that even FA still maintains that AA may not be FS. But as the three independent DNA testing show that AA is FS for 99.99% to 99.99999%, the burden of proof is on the side of those who question the DNA testing. Unless you have evidence, you have to accept the hypothesis at least tentatively. That's how the society works.

Dave,

Would you please post the source of these statistics again (99.99% to 99.99999%)? Thanks.

[helenazar]

... the burden of proof is on the side of those who question the DNA testing. Unless you have evidence, you have to accept the hypothesis at least tentatively. That's how the society works.

Dave,

I tried to explain this concept previously on other threads, but I don't think I ever got this point across and only succeeded in getting people upset. The idea has always been that the new hypothesis is what must be proven - not the old one disproven - this is not a new idea by any means.

But people often seem to get upset when you try to explain this to them and say that whoever says this, is trying to push his own opinion and doesn't want to accept any others. I never could understand why it's such a controversial issue here, since this is the rule by which the rest of the civilized world generally operates...  

[daveK]

Good point, Helen.

Gill explained his 1/300 odds in his paper in 1995.



However, that data was outdated, and didn't incorporate the Bayesian method. It's not only mtDNA. How many people do you think there were a girl who looked like Anna Anderson, almost same age, same hair/eye color, who disappeared in a same time when AA appeared in the same region!! You have to take that into account by Bayesian inference (the court of law requires forensic scientists to use Bayesian). Here is my calculation. Come to think of it, I will also post the population genetics paper I cited in this calculation.

Question 2) Is  the random match probability of AA’s DNA really 1/300?

Some AA proponents assert that AA’s specific mtDNA type is very common type, therefore a match between AA and FS is just by accident. However, this argument is fundamentally flawed. If so, why don’t they just show the data of someone who has same mtDNA? There are more than dozens populaiton genetics papers that you can check very easily. They can’t, because their claim is not true.

Before showing the evidence,  I have to point out that the probability 1/300 reported in Peter Gill’s study in 1995 was outdated. Gill “guessed” the number from statistical average because he didn’t find AA’s mtDNA type in database available in 1995. Therefore, any unknown mtDNA in 1995 was estimated as “1/300” temporally, even if its actual probability is 1/5000 or 1/100,000 (!).  

To get more accurate estimate, I checked all mtDNA (HVI) database available to me that contained 8,902 sequences of European Caucasian including US Caucasian, British, French, German, Italian, Spanish, Polish, Russian, Hungarian, Austrian, Dutch, Norwegian, Swedish, Ashkenazic Jewish, Belgian, Icelandic, Austrian, Bulgarian, Portuguese and so on. I also checked African and Asian population just in case. Most convenient sources are major human genetics journals such as Annals of Human Genetics and American Journal of Human Genetics (especially Annals of Human Genetics vol 67 (2003), p281 was helpful). Also computerized database were used, such as NCBI GenBank, European Molecular Biology Laboratory (EMBL), and US Department of Justice FBI CODIS database.

The reason why I investigated different regions separately was to see “population structure” due to ethnic subgroup, but prevalence of Tara clan was 10 +/- 2% in all countries in Europe, which indicates there is no siginificant structure (also see Science Vol 254 p1735). I’ll discuss this issue in Question 3.

TABLE 4 (Some examples of European mtDNA (HVI) studies)
---------------------------------------------------------------------
French (total = 109)
9 person has the most common  type: CRS (no mutation)  
Almost all other 93 person has a unique mtDNA (does not share mtDNA each other).
No one has AA’s mtDNA  (16126C, 16266T, 16294T, 16304C)
----------------------------------------------------------------------
Autstrian (total = 101)
9 person has the most common  type: CRS (no mutation)  
Almost all other 80 person has a unique mtDNA (does not share mtDNA each other).
No one has AA’s mtDNA  
----------------------------------------------------------------------
British (total = 100)
12 person has the most common  type: CRS (no mutation)  
No one has AA’s mtDNA  
-----------------------------------------------------------------------
Russians and Ukrainians (total = 201)
22 person has the most common  type: CRS (no mutation)  
No one has AA’s mtDNA  
-----------------------------------------------------------------------
Polish (total = 436)
67 person has the most common  type: CRS (no mutation)  
No one has AA’s mtDNA
-----------------------------------------------------------------------
US Caucasians total = 323
61 person has the most common  type: CRS (no mutation)  
No one has AA’s mtDNA


In all regions, by far the most common mtDNA haplotype (HVI) is CRS (Cambridge Reference sequence). About 10% of population in any country (except US) has this sequence (almost same prevalence as AB blood type), i.e. about 65 million European has an exactly same mtDNA sequence (at HVI). There is no known reason why this specific type is so prevalent. It seems just stochastic genetic drift event. A friend of mine jokes this mtDNA type is related to “beauty phenotype” expressed in their daughters, but I don’t think it’s true. (By the way, this CRS sequence itself from a British woman whose identity kept secret for some reason since 1981. A rumor goes that it was a researcher’s wife’s mtDNA.)

However, this CRS mtDNA is an exception. Almost all other mtDNA type is rare, usually less than 1%. For example, I checked Tsarina’s mtDNA type 16111T/16357C. There was 0 in database of 8902 caucasians. Tsar’s mtDNA was also rare, 0 out of 8902. And Anna Anderson’s mtDNA had 1 in 8902 (1 found in Iceland study). therefore the random match probability is  1/8902 = 0.01%: about 30 times rarer than the original Peter Gill’s estimate (1/300).

So, can I conclude from this DNA evidence alone? Not so fast. I think many people confuse DNA’s random match probability, likelihood ratio, with Posterior Odds. To discuss if AA is FS, we have to discuss posterior odds.

Bayesian inference is the logical/mathematical framework to interpret the combined probability of independent event. Forensic science in both US and UK are always interepreted in a logical sturucture of Bayesian inference. In the court, forensic exprert are instructed by judge to testify only regarding to “DNA random match probability” or “likelihood ratio”, but what really concern jury is the posterior odds. Here I try to be a jury rather than a DNA expert.

O (posterior) = O (prior) * DNA likelihood ratio

Roughly speaking, if two person’s sex, age, physical feature including height, hair color, face feature, prior odds are 1:10. Considering FS has been missing at almost exactly same time at same geological area as AA appeared, even conservative odds brings this to 1:100. DNA random probability is a simply inverse of likelihood ratio in this case, so my calculation shows:

O (posterior) = 1/100 x 1/9000 = 1/900,000 (that is to say, probability that AA is FS is 99.9999%)

As “reasonable doubt” is generally considered O(posterior)(threshold)
=1/10,000, it is reasonable to accept hypothesis that “AA is FS”.

Therefore, with overwhelming evidential support and lack of alternative scenario, I support the hypothesis that AA= FS.

[daveK]

I have to note: My calculation is a sort of new hypothesis. It was never scrutinized by experts, or published in a peer-reviewed journal. Therefore, the burden of proof is on my side. If you doubt my calculation, you should trust P Gill's original calculation.  

[helenazar]

My calculation is a sort of new hypothesis. It was never scrutinized by experts, or published in a peer-reviewed journal. Therefore, the burden of proof is on my side.

This is why I wanted to know exactly how you arrived at these numbers. Sounds reasonable, but a statistician needs to evaluate them in order to validate them. Until that happens we should probably go by Peter Gill's very conservative odds: 300 to 1 in favor of AA=FS.

[daveK]

If FS was murdered by Grossmann  in 1920 then she could not have been AA.......

All the facts are not in.  We hope that Penny will one day return and tell us their findings.  If not,  then we'll have to get the results from their Atlantis.  Until then,  I'm still outside the box and unwilling to jump into the box with AA being FS.

AGRBear

Bear, I am not questioning your understanding of DNA. And it is basically a great thing to explore any alternative scenario, by thinking everything critically. Your theory is very interesting.

But all I am saying is that the burden of proof is on your side if you want to claim that FS was murdered by him. The whole argument is based on your conjecture. You have to provide us the authentic literature as a source. that's the basic. That’s the whole point of this thread.

And you cannot rely your thesis on the “future” literature.

I could say that George W has an identical twin who also serves a President and I know someone who will publish the evidence “in the future”. But I cannot say the evidence because that is a secret until it is published. Could you believe me?

[AGRBear]

Let me say, again, it doesn't matter to me if AA was FS.   Until I came to this forum,  I didn't even read very much about AA.  I had a few books but I never read them, just glanced at them, read a few pages here and there....  Still haven't.  However, I don't like conclusions being made when not all the facts are in for us to see, think about and make our own conclusions.

The reason I have a problem with the results of the mtDNA proving AA was FS is two fold:

(1) If FS was murdered by Grossmann  in 1920 then she could not have been AA.  

All the DNA testing can not make AA  iinto being  FS who might have been dead since 1920.

(2) The test, which shows AA's mtDNA  and  Karl Maucher as a relative tells us only that, they were relatives.  It means AA was related and this range, I'm told, can be within 10 to 25 generations.

There apparently is no  difficulty in saying Karl Maucher is a realtive of AA  but this does not mean AA is a relative of FS.  Because,  it may be possible that FS and Gertrude, the grandmother of Karl Maucher, may have had different mothers.  And mtDNA is a test proven through the maternal lines.

The tests are best in  proving  AA was not GD Anastasia than it does showing us that AA was FS.

Don't jump all over this post telling me I don't understand DNA or mtDNA.  Helen has done an excellent job of explaining it to me over on the DNA thread as well as in PM.


SOOOOOoooooo,  all the facts are not in.

We hope that Penny Wilson will one day return and tell us their findings about FS and Gertrude.    [If not,  then we'll have to get the results from her and Greg from their Atlantis or a new book]. Until then,  I'm still outside the box and unwilling to jump into the box with AA being FS.

AGRBear

[daveK]

This is why I wanted to know exactly how you arrived at these numbers. Sounds reasonable, but a statistician needs to evaluate them in order to validate them. Until that happens we should probably go by Peter Gill's very conservative odds: 300 to 1 in favor of AA=FS.

No Helen, it doesn't require statistician to validate the number. IMHO, it is intellectually lazy to stick to the obviously outdated number. In 1995, there were many economists who predicted the Dow Jones Industrial Average would be 15,000 by now. If that was true, I wouldn't be writing this with this old computer. We have to update the number. I will post more original data to convince the people to accept new number.

[daveK]

Let me say, again, it doesn't matter to me if AA was FS.   Until I came to this forum,  I didn't even read very much about AA.  I had a few books but I never read them, just glanced at them, read a few pages here and there....  Still haven't.  However, I don't like conclusions being made when not all the facts are in for us to see, think about and make our own conclusions.

The reason I have a problem with the results of the mtDNA proving AA was FS is two fold:

(1) If FS was murdered by Grossmann  in 1920 then she could not have been AA.  

Bear, I am not denying that there is a possibility that FS was killed by Grossmann. I am talking about the RELATIVE PROBABILITY of the event. No one would be ever sure that AA was FS, except AA herself, or even AA herself might have not been so sure at old age.

Here is my conjecture:
Anna Anderson was FS = 99.9999%
Anna Anderson was Anastasia = 0.00000000 (add 80 of zero here)0001% *
FS was murdered by Grossmann = 0.00001%
FS was murdered by other murders = 0.00002%
FS was killed by accident = 0.00002%
FS was living peacefully under other pseudonym = 0.00004%
FS was kidnapped by foreign intelligence agency such as KGB = 0.00001%
FS didn’t exist from beginning, she was a fiction by her family= 0.000001%

If we talk about any small possibility, it goes ad infinitum…

[daveK]

However, I don't like conclusions being made when not all the facts are in for us to see, think about and make our own conclusions.

Also, we are NOT MAKING our CONCLUSION. We are trying to reach MOST PROBABLE HYPOTHESIS.

You say "when all the facts are in", but there NEVER be the time when all the facts are in. Never.

[daveK]

We hope that Penny Wilson will one day return and tell us their findings about FS and Gertrude.    [If not,  then we'll have to get the results from her and Greg from their Atlantis or a new book]. Until then,  I'm still outside the box and unwilling to jump into the box with AA being FS.

AGRBear

Bear, if you have origianl information from Atlantis against AA=FS, i would appreciate if you could post the pictures or type the text. I will also try to find it from library. I may have to adjust my estimante of 0.00001% depending on the context.

[Annie]

Bear, I am not denying that there is a possibility that FS was killed by Grossmann. I am talking about the RELATIVE PROBABILITY of the event. No one would be ever sure that AA was FS, except AA herself, or even AA herself might have not been so sure at old age.

Here is my conjecture:
Anna Anderson was FS = 99.9999%
Anna Anderson was Anastasia = 0.00000000 (add 80 of zero here)0001% *
FS was murdered by Grossmann = 0.00001%
FS was murdered by other murders = 0.00002%
FS was killed by accident = 0.00002%
FS was living peacefully under other pseudonym = 0.00004%
FS was kidnapped by foreign intelligence agency such as KGB = 0.00001%
FS didn’t exist from beginning, she was a fiction by her family= 0.000001%

If we talk about any small possibility, it goes ad infinitum…

Exactly. And IF I had died when that truck ran me off the road back in 1984, I wouldn't be here.

There really isn't much evidence other than pure speculation that she MAY have been a victim. There is a lot more on the side of her living and being AA.

Even I didn't know it was 99.9999% until DaveK posted it, I thought it was about 75% (which was still enough for me.) But 99.9999%, well, come on. That's as good as it's going to get. Even the dumb hicks on Maury Povich who get those results on paternity tests don't still question 'hey if there's a 00.0001% chance it ain't my kid I ain't payin'!' I have even seen cases where the guys got a 96% of being the father and still accepted it! Those results are overwhelming, and since there is no pure 100% in DNA tests that really is the end of the line.

[Alice]

What a fascinating thread! It's taken me an hour or so to thoroughly read through everything on this thread, but it's been worth it.

Thankyou so much, DaveK.

And Annie . . . unfortunately, some people will never accept the 99.9999% probability that AA was FS, and the 0.0(+ 80 zeros)01% that AA was not AN. Sad, but true.