Wednesday 19 June 2013

BritainsDNA, The Times and Prince William – the perils of publication by press release

The news of Prince William’s Indian DNA was featured in a number of newspapers around the world on Friday. The Daily Mail reported that “DNA testing has revealed that Prince William will become the first British monarch of Indian ancestry”. The Times clearly considered the story so momentous that it devoted most of its front page to the news and also carried a huge spread across pages four and five. Prince William was also the subject of a somewhat gushing editorial marvelling at “the power of science in the age of genetics to illuminate our history” and extolling the virtues of genetic genealogy which “can solve personal mysteries by linking individuals to vast extended families they did not know they had”.1

The Times is normally regarded as one of the more serious "quality" newspapers in the UK. It was therefore surprising to see a seemingly trivial story more suited to the pages of a family history magazine receiving such prominent and extensive coverage. Furthermore, the story was prominently linked to a readers' offer for the commercial genetic ancestry company BritainsDNA (which also trades as ScotlandsDNA, IrelandsDNA and YorkshiresDNA), who had very conveniently arranged for their new Chromo 2 test to be launched to coincide with the publication of The Times feature. A large box that was to all intents and purposes part of the editorial content on page four of The Times invited readers to discover their "ancient history" with BritainsDNA. As a Times+ subscriber I was advised that I could order a "free upgrade package worth £65" if I ordered a test from BritainsDNA.

The DNA tests for the research were said to have been carried out by Jim Wilson, "a genetics expert at the University of Edinburgh and BritainsDNA". The Times did not clarify that Jim Wilson is a Director of the Moffat Partnership, the name by which BritainsDNA is registered at Companies House, and that he therefore has a direct financial interest in the company.2 It is also not clear whether the research was done by Jim Wilson in his capacity as a director of BritainsDNA or as part of his job as Senior Lecturer in Population and Disease Genetics at the University of Edinburgh. Somewhat surprisingly, The Times made no attempt to seek comments on the story from any other expert geneticists for the sake of balance. Nor indeed did any of the other newspapers.

The results of scientific research are usually published in peer-reviewed scientific journals. The peer-review process is particularly important for science because it helps to filter out unreliable research and it also ensures that the full methodology and supplementary information are published with the paper so that the results can be scrutinised and verified by other researchers. The Times makes no mention of a forthcoming paper in a scientific journal, and it would therefore appear that the research is based solely on information given to The Times by BritainsDNA.  The fact that the researchers have bypassed the peer review process and published their results in such an unorthodox way means that there are many questions about the research which have not been answered. The company have a history of making claims in the media and to date none of their research has appeared in a scientific journal. This affair also raises serious questions about the judgement of The Times who have effectively provided an extended advertorial for a commercial company purely on the strength of research that has only been published in a press release.

mtDNA results
The royal ancestor at the centre of the Indian DNA story is a lady by the name of Eliza Kewark, who is the great-great-great-great-great-grandmother of Prince William and Prince Harry. Eliza lived in Surat in the Indian state of Gujarat, but she is usually described in family records as Armenian. To determine the mitochondrial DNA haplogroup (branch of the human mtDNA tree) of Eliza Kewark mtDNA testing was done on two of Prince William's cousins who are direct descendants of Eliza on the direct maternal line. mtDNA is passed on by a mother to her children, but while males inherit their mtDNA from their mother they cannot pass it on to the next generation. As Princes William and Harry and their mother Princess Diana are also direct matriline descendants of Eliza Kewark it can therefore be safely inferred that they would share the same mtDNA haplogroup as their cousins, provided of course that the genealogical research is correct and that there were no hidden adoptions in the family tree.

The Times article does not reveal the level of mitochondrial DNA testing that was performed on Prince William's cousins. However, as BritainsDNA are claiming credit for the research we can only assume that they used their own chip for the tests. The standard BritainsDNA mtDNA test looks at around 300 mtDNA SNPs (markers), while their newly launched Chromo 2 test covers just over 3000 SNPs. mtDNA researchers now prefer to test the full mtDNA genome (all 16569 base pairs). While a well selected panel of mtDNA SNPs is usually sufficient to confirm the base haplogroup we would be more confident of the haplogroup prediction if the results had been made publicly available so that they could be independently verified.

BritainsDNA have confidently claimed their research "confirms… that the mtDNA of Eliza Kewark of Surat was of Indian heritage"3 which led The Times to proclaim that "the Duke of Cambridge will be Britain’s first king to have proven Indian ancestry". An examination of the evidence in the public domain reveals that this claim is something of an exaggeration.

Eliza Kewark’s inferred mtDNA haplogroup is R30b, which is said be incredibly rare, and is certainly not a haplogroup that one would expect to find in a person of British origin. However, this finding is consistent with the genealogical research, and a non-British haplogroup is only to be expected. The question then arises as to whether the haplogroup is one which is indicative of Indian or Armenian ancestry. According to The Times: “Comparison with databases of more than 65,000 individuals from around the world showed that only 14 examples of R30b have been reported, 13 of whom were Indian and one Nepalese. The other related branches, R30a and R30*, are also entirely South Asian.” The databases that were searched have not been revealed but I would imagine that searches have been done in Genbank, mtDNACommunity and Mitosearch. All these public databases are heavily biased and are dominated by people of Western European origin, who tend to be the most active genetic genealogists. GenBank, for example, currently has around 18,000 full mtDNA sequences, with almost 40% of these having been provided by customers of the commercial genetic genealogy company Family Tree DNA.

Clearly, to test the hypothesis that Eliza’s maternal line is from India and not from Armenia it would be necessary to test a sufficiently large sample of people of documented Armenian and Indian origin to determine with reasonable confidence that R30b is only found in India and is absent in Armenia. As far as can be ascertained from the scant details available such research does not appear to have been done. The situation is further complicated by the fact that the Armenians have had colonies in India since the sixteenth century4 with one such colony, as reported in The Times, having been in Surat:
"There had been a thriving Armenian trading community in Surat since the 17th century and there was some inter-marriage with the local population. When Theodore [Forbes] met Eliza it was common for British men to begin relationships with Indian women and to have children, even if they already had a family back in Britain.

Susan Harvard, who has been researching Eliza and Theodore’s relationship for almost 30 years, believes that the couple were married in the Armenian church in Surat in early 1812, shortly before he was posted to Mocha in current-day Yemen. However, it may not have been legally recognised.”
There is very little in the scientific literature on haplogroup R30b. It was first discovered by Palanichamy et al in 2004.5 Chaubey et al surveyed around 12,000 mtDNAs from across India as part of an investigation into the phylogeography of mtDNA haplogroup R7 in the Indian peninsula and briefly discussed R30:
"Haplogroup R30 splits into two subclades R30a and R30b, the former supported by ten coding region substitutions and the latter by 24 coding and control region mutations. Similarly, in haplogroup R31 a new subclade R31a can be distinguished by 17 control and coding region mutations. Coalescent estimates suggest an ancient branching pattern in hgs R30 and R31, dating back almost to the earliest diversification of the superhaplogroup R itself. This most probably occurred soon after the out of Africa dispersals into the Indian subcontinent."6
I have only been able to find one paper which sampled mtDNA from Armenia. Harutyunyan, Khudoyan and Yepiskoposyan took samples from 741 self-identified Armenian males in four regional areas of the country. A further 94 samples were collected from the Armenian community of London.7 However, the researchers do not appear to have published the sequences or the haplogroups. They only sequenced the hypervariable region and did not test any coding region SNPs. It is common to find people with identical HVR1 sequences who are in completely different haplogroups. (Note, however, that population geneticists are asking different questions from genetic genealogists and for their purposes it is the relationship between the tree and population history which is important rather than a well-resolved mtDNA genealogy.)

BritainsDNA claim that R30b is rare in India and that only 0.3% of Indian people have this "lineage", though the source of this statement is not given. The fact that R30b is rare even in India makes the need for comprehensive sampling even more important. India has a population of around 1.2 billion and Armenia has a population of over three million, yet we have just a few thousand samples to represent the potentially huge genetic diversity of India, and just a few hundred samples to represent the population of Armenia. If one were to claim at least 10:1 support in favour of an Indian versus an Armenian origin then it would be necessary to test thousands of Armenians to establish that the frequency of R30b in Armenia is significantly less than 0.03%.

The BritainsDNA researchers have brought to light some interesting examples of a rare mtDNA haplogroup. However, there is no proof that Eliza Kewark's maternal line is from India, Armenia or indeed any other country. It is clear that R30b has a very ancient origin and it could, therefore, potentially have a very wide geographical spread. As we have seen there are simply not enough samples available from these regions of the world to draw any firm conclusions.

Autosomal DNA results
The BritainsDNA researchers did further tests on Prince William's cousins using autosomal DNA. These tests show that the cousins were "about 0.3 per cent and 0.8 per cent South Asian, made up of three different blocks of South Asian DNA in each of their genomes, while all the rest was European in origin". They claim that "This provides independent genetic evidence corroborating the mtDNA work" but the evidence is in fact very weak.

There is little information provided about the methodology. As BritainsDNA are claiming the credit for doing the testing I would have thought that they would have used their own chip for the tests. Their new "All my Ancestry" test uses 250,000 autosomal SNPs. Yet, in an article entitled "A royal revelation" on the BritainsDNA website we are told that the researchers read "over 700,000 markers scattered across the genome of Princess Diana's matrilineal cousins". The technical details therefore remain a mystery. We don't know which reference populations were used and which software tools were deployed to calculate the ethnicity percentages. It's common to get conflicting DNA ethnicity results when testing with different companies. However, the experience from the commercial autosomal tests on the market suggests that they are usually adequate to detect inter-continental differences so we can perhaps give BritainsDNA the benefit of the doubt and assume that the South Asian ancestry detected is genuine and not noise.

We inherit a random mix of autosomal DNA from all our ancestors and it is not easy to determine which ancestor contributed which particular segment of DNA. It can be done by a combination of phasing and triangulation if you test enough family members, but it is a very difficult, expensive and labour-intensive process. As Razib Khan has pointed out, it is in fact quite likely that Prince William has no Indian ancestry at all on his autosomal genome.8 Genetic ancestors begin to drop off our DNA tree very rapidly and the chances of inheriting a segment of DNA from a specific great-great-great-great-great-grandparent are very slim indeed. The problem has been nicely summarised by Luke Jostins:
The probability of having DNA from all of your genealogical ancestors at a particular generation becomes vanishingly small very rapidly; there is a 99.6% chance that you will have DNA from all of your 16 great-great grandparents, only a 54% of sharing DNA with all 32 of your G-G-G grandparents, and a 0.01% chance for your 64 G-G-G-G grandparents.9
It is also important to research the family trees of the two cousins on all their family lines to help to eliminate the possibility that the Asian percentages might have come from ancestors on another part of the tree. Many British people have ancestors who lived and worked in India, and it's quite possible that some people will have multiple ancestors of Indian origin on their tree. As BritainsDNA have not published the details of their genealogical research we do not know if this possibility has been ruled out.

Far from corroborating the mtDNA research the limited autosomal DNA results that have been released to the public provide more questions than answers.

Public invasion of genetic privacy?
Other commentators have cautioned about the ethical implications of revealing the mtDNA haplogroup of a public figure without consent.10, 11, 12, 13 This was not something that had occurred to me when I read the story and I am still undecided on the issue. While the haplogroup does not reveal any personal information it is possible that some people will receive DNA results that challenge their views of their identity. It is only common courtesy to ask permission before publishing information about living people, whether they are a celebrity, a member of the Royal Family or just an ordinary individual. It is a debate that we clearly should be having, though a wide consultation would be needed to reach some consensus. The situation is only likely to become more complicated in the future when full mitochondrial sequence testing becomes the norm for genealogy purposes. The full sequence test can sometimes reveal medically informative mutations, and it would be highly unethical to reveal such associations by inference and without consent.

Conclusion
In the last few days we have seen an uncomfortable blurring of the boundaries between academic research and a commercial venture. A genetic ancestry company which claims to be “closely involved in scientific research” has once again failed to follow the standard scientific protocols and has published its research in a press release rather than a peer-reviewed journal. A respected UK national newspaper has compromised its integrity by publishing an advertorial in support of a commercial PR campaign.14, 15 We have learnt that Princess Diana, Prince William and Prince Harry all belong to the rare R30b mtDNA haplogroup. This haplogroup might or might not be associated with India. Was it worth it? What do you think?

See also
Genetic astrology: a tale of misleading claims and legal threats

Related blog posts
BritainsDNA, the BBC and Eddie Izzard
Alistair Moffat, BritainsDNA and the BBC - a "uniquely British farce"
More pseudoscience from Alistair Moffat
- The British: A genetic muddle by Alistair Moffat

References
1. Articles on The Times website are available to subscribers only. The four articles on Prince William’s DNA that appeared in The Times on 14 June 2013 are:
* David Brown. Revealed the Indian Ancestry of William, p1.
* David Brown. Doomed Indian love story had happy end for the daughter who sailed to a better life, p4.
* David Brown. How a retired writer found out that he was within spitting distance of royalty, p5
* Editorial. Accidents of birth: the extraordinary tale of mtDNA haplogroup R30b has lessons for us all, p28
2. Records at Companies House show that Dr James Wilson was appointed as a director of The Moffat Partnership on 12 December 2012.
3. "Revealed the Indian ancestry of William". BritainsDNA press release, 14 June 2013.
4. Zenian D. The Armenians of India. Armenian General Benevolent Union News Magazine, 1 July 2001.
5. Palanichamy M, Sun C, Agrawal S et al. Phylogeny of mitochondrial DNA macrohaplogroup N in India, based on complete sequencing: implications for the peopling of South Asia. American Journal of Human Genetics 2004; 75(6): 966–978.
6. Chaubey G, Karmin M, Metspalu E et al. Phylogeography of mtDNA haplogroup R7 in the Indian peninsula. BMC Evolutionary Biology 2008; 8: 227.
7. Harutyunyan A, Khudoyan A, and Yepiskoposyan L. Patrilocality and recent migrations have little impact on shaping patterns of genetic structure of the Armenian population. Genetika (Russian Journal of Genetics) 2009; 45(8):1125-31.
8. Khan R. Prince William may have little to no Indian ancestryGene Expression blog, 14 June 2013.
9. Jostins L. How many ancestors share our DNA? Genetic Inference blog, 11 November 2009.
10. Middleton A. Attention The Times: Prince William's DNA is not a toy. The Conversation, 14 June 2013.
11. Hern A. Are there ethical lapses in the Times' story on William's "Indian ancestry"? New Statesman, 14 June 2013.
12. Brice P. Public invasion of genetic privacy for UK royal family? PHG Foundation, 14 June 2013
13. Khan R. Ancestry should not be subject to privacy restraints. Gene Expression blog, 14 June 2013.
14. Johnson B. How a spit of royal DNA makes money for Rupert Murdoch: When science in the news goes terribly, terribly wrongStories that Matter, 14 June 2013.
15. Greenslade R. The Times's Prince William splash linked to readers' offerGreenslade blog, The Guardian, 14 June 2013.

Resources
The ISOGG Wiki has charts comparing the services of all the major genetic genealogy testing companies. The Wiki also includes a list of genetic genealogy mailing lists and forums where people can ask for advice about DNA testing and understanding their DNA results.

Sense About Science have provided a pamphlet entitled Sense About Genetic Ancestry Testing which highlights the limitations of the inferences that can be made from deep ancestry tests.

My commissioned blog post for Sense About Science entitled Sense About Genealogical DNA Testing explains the practical applications of DNA testing for the family historian.

Exploring your Indian ancestry
If you have ancestors who lived in British India I recommend that you join the Families in British India Society. They have an excellent website with a huge selection of genealogical sources which can be found at www.new.fibis.org and they offer many additional services for their members. FIBIS also have their own DNA project which is open to both members and non-members. Further details can be found at www.new.fibis.org/dna.

© 2013-2015 Debbie Kennett

14 comments:

DNACousins said...

I was actually pleasantly surprised by the level of detail given in a popular newspaper article. The paper trail, the rarity of the mtDNA haplogroup, the autosomal DNA, and the known presence of an Armenian community in India all form a coherent picture with a "preponderance of evidence" for the conclusion. Certainly people who are already deeply involved with genetic genealogy would be curious about the exact markers used, but this is a human interest story that would be beyond the scope of most technical journals.

The 3000+ mtDNA markers used to assign haplogroups are in the ballpark of chip-based technology used by 23andMe and National Geographic's Genochip (both of which created their own SNP selection but neither of which include the specific marker for R30b). The 250K autosomal markers are in the ballpark of SNPs carried over from Illumina's 550K chip to the 700K chip. The 550K chip was tested in the Human Genome Diversity Project panel, which is the foundation of virtually all biogeographic analyses performed by other testing companies, such as 23andMe, AncestryDNA, and Family Tree DNA. The issue of privacy is confounded by the fact that it was two other lines of descent that held the key to the discovery, and those people clearly gave their consent.

Anonymous said...

Testing distant relatives is questionable science if you are trying to determine something about a living person's heritage. If they tested my distant cousins, they could easily come up with a vastly different result about my projected ancestry than if they test me. I have seen way too many bad judgment calls along this line when used in other cases. They assume that person A and B were connected because a couple of distant descendants had some interconnected DNA, but they ignore the possibility that one or both potential ancestors had siblings so the siblings could be the direct match and not person A or B.

I am surprised they didn't snatch some of the Prince's DNA from a rubbish pile and test it directly.

It's a shame the company and media chose to be less than forthcoming on revealing specifics.

Debbie Kennett said...

It was indeed encouraging to see so much detail given in a newspaper article, but this still begs the question as to why a serious quality newspaper devoted such an unprecedented amount of space to the story. It would have been more appropriate to have the article in a supplement rather than in the main newspaper. The research would probably not be of interest to a genetics journal but it still should have been published somewhere before being released to the press. The Society of Genealogists publishes a quarterly journal Genealogists' Magazine which often has scholarly articles on the genealogy of the royal family, prime ministers, etc, complete with full source citations. There are many scholarly and popular history journals which would probably be happy to publish such a story.

The preponderance of evidence does point to Indian ancestry, but that is not the same as proof. Scientists are expected to follow rigorous standards and to be transparent about their methodology. We shouldn't be left to speculate.

I'm sure the actual testing itself is sound. The company are probably using a similar Illumina chip to the other companies that has perhaps been customised with their own Y-SNPs. 23andMe and Family Tree DNA both provide technical details on their websites of the chips that they use. BritainsDNA should do the same.

Debbie Kennett said...

Close family members can indeed get very different admixture results and it's not really possible to draw too many inferences from autosomal DNA tests of distant relatives. However, as has been done here, it is possible to confirm relationships by testing the mtDNA of female relatives as a proxy.

It is actually illegal in the UK to "steal" someone's DNA and have their DNA tested without their consent:

http://www.hta.gov.uk/legislationpoliciesandcodesofpractice/legislation/humantissueact.cfm

However, DNA theft is not illegal in many other countries. There was a disgraceful case a few years ago where some Belgian researchers stalked the descendants of Adolf Hitler and extracted DNA from cigarette butt and napkins:

http://www.dailymail.co.uk/news/article-1213589/Hitlers-secret-family-How-researchers-tracked-39-living-descendants-dictator.html

http://www.dailymail.co.uk/news/article-1305414/Hitler-descended-Jews-Africans-DNA-tests-reveal.html

The Royal Family and celebrities could be particularly prone to such dodgy practices, but newspapers have a moral obligation not to publicise such unethical research.

pconroy said...

I don't know about frequencies of R30b, but looking at my 1,650+ 23andMe Results, I see one "cousin" who is R30a and is a South African, living in Israel.

I have a number of relatives who are "Cape Colored" in South Africa, and so a 3-way mix of European, African and Indian. So R30a is likely of Indian provenance here.

Debbie Kennett said...

Paul, Thanks for sharing that. With so few R30 samples available it's very difficult at present to draw any firm conclusions. We desperately need more people from the Indian subcontinent to participate in DNA testing.

Anonymous said...

What is the position of Edinburgh University in this matter? Have they ever commented?

Debbie Kennett said...

I've not seen any comments from the University of Edinburgh. They do not have any press releases about the story on their website.

Note too that Alistair Moffat, the founder of BritainsDNA, who is also the Rector of St Andrews University, has been asked to "delineate his University and personal business":

http://www.nature.com/news/the-right-to-speak-out-1.12758

http://www.thesaint-online.com/2013/04/university-slams-moffat-for-stifling-debate/

Anonymous said...

It is not so much if they commented on story X or Y. The question is more whether their rules of "good scientific practice" are observed or not. If their name is linked publicly to data I would expect that they have a requirement that such data is peer reviewed prior to publication.

Debbie Kennett said...

The University of Edinburgh offer advice to their staff on handling the media:

http://www.ed.ac.uk/schools-departments/communications-marketing/press-and-pr/staff-support

Here are their codes of practice for research staff:

http://www.ed.ac.uk/schools-departments/institute-academic-development/research-roles/research-only-staff/advice/codes/research-code

Anonymous said...

It was interesting to find this blog. I've just been dna tested and I have R30b haplogroup. My ancestry is almos totally North indian with a small percentage of south indian and nothing else.

Debbie Kennett said...

Hello RB. Thanks for sharing your results. Have you taken a full mitochondrial sequence test? If so, have you considered uploading your results to the GenBank database to make them available for research purposes? You can find details here:

https://isogg.org/wiki/GenBank

Cbg96 said...

I have distant Anglo-Indian ancestry and I have a few matches of Armenian descent on 23andme and Gedmatch. I don't show any West Asian DNA on 23andme though, just Southern Indian. It looks like Armenians may have mixed with local Indian people, so it's possible that Eliza Kewark was of Indian and Armenian descent. Vivien Leigh is also said to have been of Armenian descent.

Debbie Kennett said...

Thanks Cbg96 for sharing your experience. Eliza Kewark may well have been of Indian and Armenian descent. Unfortunately the research from BritainsDNA didn't really answer any of the questions they set out to investigate.