Sunday, 25 July 2010

Genotyping journalists

Take a bunch of journalists and ask them to undergo genetic testing - the outcome was always going to be entertaining. This was the part of Friday’s inaugural UK Conference of Science Journalists, held in the opulent surroundings of London’s Royal Society, that I was most looking forward to.

The public testing was part of the Future of Genomics session, organised by Zoe McDougall from the sequencing technology company Oxford Nanopore Technologies who provided sponsorship.

The testees to be paraded as genomic curiosities were Ed Yong, Martin Robbins, Claire Ainsworth and Mark Henderson. Three broad themes emerged from the resulting spectacle.

Firstly genotyping is confusing, no matter how impressive your scientific credentials. Claire Ainsworth has a PhD in fly genetics, Ed Yong writes a renowned biology based science blog, and Mark Henderson knows at least 50 genetic ideas that you really should too. Yet the panellists were all bemused or overwhelmed by at least some of their data. This is not criticism of their expertise but an indication of the complexity of the information involved.

Mark was genotyped by 23andMe, deCODEme and Pathway and has written in detail about his experience (subscribers to Times online content can read his three blog posts). The tests identified an increased risk of exfoliation glaucoma, which he knew from family history, but the companies differed in their estimate of background risk, leaving Mark perplexed:

How prevalent is exfoliation glaucoma, and how much do I need to worry about it? If 23andMe is correct, then even though I have a high risk genetic variant, my overall risk of this condition remains low. If deCODEme is right, I have a one in three lifetime risk. And if Pathway is right, the population risk is low, but the peculiarities of my genome dramatically enhance my chances of getting it.... I am now a confused customer!

Ed, who took the 23andMe test, has also blogged about the experience. He initially took his cystic fibrosis big fat negative at face value, but a genetic councillor quickly got in touch to explain the results were less clear cut. The test was for the most common cystic fibrosis variants, but over 1000 have been identified giving the potential for false reassurance. You can follow the discussion in Ed’s post comments, and there was a swift response from 23andMe.

That Mark and Ed were willing to share their results gave them the opportunity to receive input from experts. Mark sent his data to be analysed by Daniel MacArthur, researcher from the Sanger Centre, author of the influential blog Genetic Future, and presenter of the opening talk at the session.

Daniel’s analysis (further details in Mark’s second post) showed that out of the 554,618 variants called by both 23andMe and deCODEme, there were differences in only 77, giving an error rate of 0.007%, far lower than conventional diagnostic kits on the market.

This open-access sharing and caring model is just what 23andMe encourage with their community membership and genotype comparison technology. And, as Mark and Ed found, this is a productive way of navigating the information tangle.

But the testees’ own genetic knowledge and their ready access to experts are not shared by every consumer. As Ed pointed out, you cannot fault the information that 23andMe provide, but you need time and knowledge to take it on board.

My second theme is that genotyping generates emotions as well as information.

Testee Martin Robbins, who doesn’t possess a genetics background, was more illustrative of your standard genomic punter. He described the impersonal result screen as like facing a “wall of death” and said he found the whole testing procedure uncomfortable.

His test, carried out by Navigenics, identified an increased risk of Alzheimer’s, which as Martin pointed out will be shared by at least one of his parents, neither of whom consented to be screened.

Martin also highlighted fears about data ownership and protection from hackers and third parties such as future employers and dating sites. This was raised later in the session by Alison Hall from the PHG Foundation, who discussed the conflict between consumers posting their haplotypes on Facebook sharing their data online and the fundamental principles of data security and confidentiality.

Claire was the most circumspect about the test, phoning her parents in advance to gauge their thoughts on being indirectly genotyped. She also choose not to disclose all the data, feeling the information was not hers alone to share and pointing out that today’s innocuous variants could be tomorrow’s risk magnets.

Although the testing provoked a range of emotions and there was much talk of hypothetical dangers, no one seemed to regret having taking the plunge. This is in line with the general lack of concrete evidence for damage caused by consumer genetic tests.

My final point is that the tests provided more fodder for jokes than they did undercover meaningful clinical insights. Ed’s blog post is full of smart gags (a hypothetical love child with Mark, science writer disease risk top trumps and too much talk of wet earwax). My favourite wisecracks from the session were Claire’s disappointment that her phenotype was less sexy than the dark-eyed curvy genotype prediction and Martin’s dismissal of his increased prostate cancer risk given his expected Alzheimer’s oblivion.

Ed’s results gave the starkest illustration of the insignificance of much of the potentially non-trivial information. Most genetic studies have been carried out on Europeans, so his Asian ancestry meant that his risk predictions are currently very unreliable and practically useless. But Ed now owns a copy of his genotype so will be able to uncover future associations as research becomes more geographically widespread.

There was also a genetic councillor, Christine Patch, on the panel who outlined the differences between the targeted clinical tests provided by the NHS and those of low predictive value available direct to consumers. For example, the NHS do not screen for the Alzheimer’s variants flagged in Martin’s results as the tests are not deemed clinically useful.

My only criticism of the session was that there wasn’t room on the agenda to explore the implications of full genome sequencing. James Brenton, from Cancer Research UK, hinted at the future complexity in his presentation on cancer genomics. He described research showing genetically distinct populations of cancers cells present in different locations of the same tumour, and how the populations evolve and shift in response to therapy.

Neither was their time to discuss the previous day’s US hearing on regulation of the direct-to-consumer genetic test industry. If Daniel MacArthur was tired from staying up to post his analysis of the events on Genomes Unzipped, he didn’t show it.

The rest of the day was equally stimulating, a raft of expert speakers and bubbling discussion. There was much talk about the role of bloggers, social networks and cyberspace review. If the web provides an audience for newcomers typing from their kitchen table then this conference was equally inclusive. The day was made accessible by the efforts of British Association of Science Writers members and sponsorship from Elsevier and others.

Other conference attendees received genotyping kits, not all could present on the day. I am waiting for the results of a deCODEme cancer scan which will go in a separate post, and I will update this post with any links to others who share their experiences.

For me, it was a full, blissfully child-free, day immersed in science and new communication technologies. I left with a humming mind and a curiosity about what story my DNA would tell.


  1. Thanks for an interesting post. The lady from the PHP Foundation seems somewhat confused about the difference between a haplotype and a haplogroup. A haplotype consists of the personal results from a DNA test (short-tandem repeat markers for a Y-DNA test and differences from the Cambridge Reference Sequence for an mtDNA test). A haplogroup is a broad genetic grouping of people who share a common ancestor from thousands of years ago. 70% of British men for instance belong to haplogroup R1b1b2. There are therefore no privacy implications in posting haplogroup information on Facebook as in the link given above.

  2. Thanks for the information Debbie.

    The error is all mine and nothing to do with Alison from PHG. Facebook and haplogroups were mentioned in the session discussion rather than by Alison directly (the misuse of “haplotypes” instead of “haplogroups” is also mine). I can see that the hyperlink I chose to illustrate the point doesn’t do it very well – I have edited the post accordingly.

    Another example of the benefit of having these discussions online where those with different expertise can contribute!