Equal opportunity genomics

ScienceDaily reports a bold strike against the human genomics patriarchy:

Geneticists of Leiden University Medical Centre (LUMC) are the first to determine the DNA sequence of a woman.

Well, there are other female genomes out there - the 1000 Genomes Project has already generated moderate-coverage sequence for at least one female (an anonymous woman of European ancestry, collected in Utah) and low-coverage sequences of perhaps half a dozen more, and I think the Cancer Genome Project has sequenced at least one tumour sample from a female patient - but perhaps Leiden beat them to it, and at the very least they can legitimately claim to be the first to realise that the gender of a genome donor by itself was enough to warrant publicity.

Public relations aside, there's nothing hugely biologically interesting about sequencing a female genome - it just means you get one extra copy of the X chromosome. As the article notes, that does give us a little bit of extra information about the genetic diversity on this exquisitely interesting chromosome, but I'm not sure to what extent this outweighs the extra information you could get from sequencing a male (Y chromosome sequence, and accurate information about the precise arrangement of variations along the X chromosome - information that is "scrambled" when you sequence two copies at once).

The article is more interesting in what it says about the sudden explosion in sequencing capacity resulting from next-generation sequencing technologies:

"The sequencing itself took about six months. Partly since it was run as a ‘side operation’ filling the empty positions on the machine while running other projects. Would such a job be done in one go, it would take just ten weeks".

That's right - they sequenced an entire human genome using just the left-over capacity of their sequencing machines, in six months. This sort of capacity certainly isn't limited to Leiden: sequencing facilities around the world are all churning out human genome equivalents every few days. Now, consider that the first human genome took the combined work of hundreds of scientists from over 18 countries for thirteen years, and you have you a taste of how rapidly sequencing technology is moving.

I wonder how many more "first" human genomes we'll be expected to applaud over the next few months. Already this year we've had the first Nobel prize-winning genome, the first African genome and the first East Asian genome announced. Next, I suppose we'll have a spate of disease genomes: the first cancerous genome of course, but also perhaps the first autistic genome, the first obese genome, the first arthritic genome...

None of them will mean much by themselves; the exciting results will come when this exploding sequencing capacity allows researchers to compare the whole genomes of thousands of disease patients with those of thousands of controls. These studies are approaching rapidly, and in addition to their powerful information they'll bring a host of technical challenges in terms of sequence generation, data storage and transfer, and interpretation. I'll be talking a lot more about the promises and challenges of next-generation sequencing over the next couple of months - stay tuned.

Subscribe to Genetic Future.

Free personal genomics!

Over at Eye on DNA, Hsien wonders about the effects of a slowing economy on the personal genomics market. Well, no matter how hard it's getting to make your mortgage repayments, you can probably still afford personal genomics if it doesn't cost you anything:

In New Jersey, meanwhile, the nonprofit Coriell Institute for Medical Research is developing a service that will test for a slate of validated genetic markers, and provide free — yes, free — information and analysis for common diseases. The institute plans to sign up 10,000 people in the next two years, and eventually enlist 100,000 people.

(From a recent piece in Wired). You can sign up here; there is a pretty extensive FAQ here. Note that you will need to physically attend an enrollment session at the Coriell Institute in New Jersey. Also, I see that Coriell is adopting the paternalistic "need to know" approach pioneered by Navigenics, and won't provide participants with any information about genetic variants that aren't "medically actionable" (e.g. incurable disease risk variants), although they will hand out information on non-disease traits like eye colour. Still, if I lived anywhere near New Jersey I'd be signing up right now rather than wasting time writing this post.

(As an aside, I wonder why Coriell is using a saliva-based method when it could be using its considerable expertise to create and store cell lines from blood - essentially generating an endless source of DNA for researchers to analyse. That seems like a missed opportunity that someone will be seriously regretting in a few years when there's no DNA left for whole-genome sequencing, or epigenome analysis, or whatever.)

If you're more ambitious, you could also sign up for (eventual) free genome sequencing via the Personal Genome Project.

Subscribe to Genetic Future.

23andMe, deCODEme and Navigenics at Cold Spring Harbor

Just in case anyone has been wondering why I've been so quiet, I'm in beautiful Cold Spring Harbor this week for the Biology of Genomes meeting. Be warned that I'll have a lot more to say about this meeting once I've recovered from the combined effects of jet-lag and the punishing schedule (last night's evening session finished at 11:30pm!), most of which will include the words "next", "generation", "sequencing" and "wow".

For now, I want to download some thoughts on this afternoon's panel discussion on direct-to-consumer genetic testing between three giants of the personal genomics industry: 23andMe's Linda Avey, deCODEme's Kari Stephansson and Navigenics' Dietrich Stephan. The session was ably chaired by the former head of the public Human Genome Project, Francis Collins.

Each of the three company representatives was given around 10 minutes to push their vision for the personal genomics industry. While all three are selling essentially the same product (a read-out of between 500,000 and 1,000,000 variable positions from your genome), each did their best to differentiate their company from the competition.

Navigenics was first off the mark, and the tone was serious: Western society is headed for a healthcare crisis, with the combination of an ageing population with an increasing frequency of lifestyle diseases like obesity and diabetes leading us inexorably towards an increased load of common diseases. Within a generation, said Stephan, we need a solution, and the only solution is early prediction and intervention.

The Navigenics message has remained on target since its launch a month ago: this is a careful, serious, completely disease-focused company. Stephan talked up the quality control on the Navigenics testing service, which includes repeating any disease-related genotype calls that fail the first time around (unlike 23andMe and deCODEme). He also emphatically stated what Navigenics wasn't: the company will never offer testing of genetic predictors of non-disease traits like height or eye colour; will never offer genetic ancestry testing; and will not offer comparisons of family members (apparently out of concerns about unexpected discoveries of non-paternity). This is, of course, an explicit attempt to portray the competitors - who both offer both ancestry, trait prediction - as frivolous.

Next off the rank was Google-funded 23andMe. Avey immediately turned Stephan's accusation back on itself: rather than being frivolous, 23andMe simply "looks at genetics holistically", with information on non-disease traits and ancestry being part of the big picture that customers want ("genealogy is the second most popular hobby on the internet," she said. "You can guess what the most popular one is.")

Avey explained that her decision to found 23andMe was based on her frustrating experiences at Perlegen attempting to recruit large cohorts of patients to use for large-scale genetics, and during her talk there was a strong emphasis on 23andMe as a system for driving genetic research. Just as the internet-based communities of Web 2.0 are willing to share their information with one another and reap the benefits, 23andMe is apparently aiming to create "Research 2.0", in which patients volunteer their genetic and clinical information for researchers to work with, and then use the results of that research to inform their own lifestyle decisions. Avey cutely terms this model "23andWe".

Finally, the intimidating Viking-like figure of Kari Stephansson took to the platform. Stephansson played continuously on the research cred of deCODEme's parent company, deCODE - not without justification, given the impressive list of large-scale genetic studies performed by the company (which Stephansson rather heavy-handedly flashed up on the screen, one by one, for what seemed like forever).

In fact, these distinguished scientific credentials seemed to be pretty much the only thing Stephansson could find to distinguish deCODEme from its rivals. Otherwise, the deCODEme ethos seems to resemble the open information model of 23andMe rather than the more old-fashioned paternalistic approach of Navigenics. "Is it always laudable when people learn more about themselves?" he asked, and argued that the answer was unambiguously yes. "Our customers will benefit from knowing themselves better."

After the three company talks, there were presentations from Johns Hopkins' Kathy Hudson and the National Coalition for Health Professional Education in Genetics' Joseph McInerney. Hudson emphasised the strong appetite of consumers for genetic information, and the desperate need for empirical data regarding the responses of people to information about genetic risk variants (which I've wondered about myself). McInerney made a strong case for the complete under-preparedness of health care providers for the boom in personal genetics, and introduced a new database called GeneFacts (currently under development) which will store expert-curated information about genetic tests for both consumers and health providers.

The following discussion was lively, and nowhere near as hostile as I expected. A few highlights:

• Stephansson was remarkably candid about the usefulness of current tests involving common variants (which all three companies rely on): in one exchange he noted that "we are marketing these tests without any claim that they will impact on people's lives"; in another, he admitted that common variants probably provide marginal utility beyond simply collecting general information on family history.

• Kathy Hudson responded to the problem of patients being given data of very limited predictive value with a very sensible solution: "In the absence of demonstrable harm, the default should be to provide the information." In her talk, she referred to the argument that genetic tests should only be ordered through a health-care provider as "an old-fashioned model" - a clear rebuff to both Navigenics, which boasts about using an in-house doctor to authorise all of its tests, and the American College of Medical Genetics, which recently issued a statement which argues that "a knowledgeable health professional should be involved in the process of ordering and interpreting a genetic test".

• Stephansson was sceptical of Avey's claims that 23andMe can perform useful research, given the limitations of self-reported data (I agree). Avey explained that this problem is one of the reasons why 23andMe is interested in working with companies like Google to integrate genetic data with medical records - a suggestion that resulted in some shocked muttering from the audience.

• Eric Lander (founding director of the Broad Institute) suggested that the medical genetics community needs to play a stronger role in the field of personal genetics, possibly contributing to some sort of expert-curated wiki-style database of information about genetic associations. He indicated that this couldn't be privately funded due to conflicts of interest. McInerney suggested that the soon-to-arrive GeneFacts database could serve as a starting point; Hudson stated that there has been substantial discussion about a genetic test registry at the Genetics & Public Policy Center, and that clear support from the medical genetics community would help this move forward. It appears that DNA Perspectives may have some competition in the very near future. Either way, this is good news for genetic test consumers.

• There was an interesting back-and-forth between Stephan and Stephansson over the issue of ancestry testing both in their talks and in the discussion: Stephan basically sees ancestry testing as a distraction from the serious business of disease, while Stephansson argued that the effect size of disease risk variants frequently varies between populations, so ancestry testing is highly relevant to calculations of disease risk. This is an empirical question that will soon be answerable (deCODE is apparently currently investing in association studies in Asian populations).

• All three of the company representatives mentioned their interest in developing whole-genome sequencing capabilities, which is unsurprising - sequencing has always been the Holy Grail of personal genomics, with the current SNP chip technology really little more than a crude place-holder until sequencing prices drop.

I also chatted with both Stephan and Avey after the session - for what it's worth, they're both extremely personable, and there was no overt animosity between any of the three competitors (somewhat to the disappointment of the audience, I suspect).

Anyway, that's the bulk of the personal genomics component of this meeting. As I hinted at the beginning of the post, the other main message has been the rapid advance in next-generation sequencing technology and its application to human genomes, which I'll hopefully be able to post about over the next few days.

Subscribe to Genetic Future.

Low technical error rates for personal genomics companies

Antonio Oliveira from Longa Vista has compared the results of his genome scans from both 23andMe and deCODEme. Of the 560,299 sites analysed by both companies, just 23 showed a different result between the two scans - a discrepancy rate of just 0.004%!

This fits with the low discrepancy rate reported by Ann Turner back in January. The take-home message: by all means worry about the interpretation of your personal genomics result, but it's likely that your actual genotype data are extremely accurate.

Via The Quantified Self.

Subscribe to Genetic Future.