In my column in today’s Guardian, I take a look at the new gene-mapping service from the Google-backed startup 23andMe. Here’s a snippet:
To provide 23andMe with a DNA sample, you have to send the firm a small test tube filled with saliva. That requirement at first struck me as distasteful, maybe even demeaning. To have one’s earthly origin and destiny reduced to a vial of spit seems like an affront to human dignity. But maybe I’m being too fussy. After all, primitive cultures have long viewed bodily fluids, and saliva in particular, as carriers of secrets about personal identity and vitality. The Cherokee tribe of what is now the southeastern United States considered saliva “a vital element,” according to the anthropologist Charles Hudson. Spittle “was to the individual as water in creeks and rivers was to the world. If one’s saliva were spoiled, it was a serious matter.”
All that 23andMe is doing is giving an information-age twist to this mystical tradition. It’s using automated laboratories, supercomputers and the internet to translate ancient symbolism into a practical digital service. Once a spiritual totem, spit is now just another informational medium.
Read.
The big story lies upstream of 23 and me:
Biotech research (for pharma, medicine, synthesis, and the human genome project) has created stunning amounts of demand for reduced cost genetic sequencing. The market has responded with many neat toys and the pipeline suggests that fairly complete $1,000 human genomes are not that many years away. So, that just “is”.
For $1,000 today, they look only at a tiny fraction of your genome. They find 800,000 or 1M places on your genome where you’re probably exactly the same as everyone else — except for one base pair (one AT, or CG pair on the ladder). Depending on which “well known locations” you are talking about, the outcome of reading that one base pair can be diagnostic for certain genotypes that have been noticed by medicine: you can (with high probability) recognize that some someone “has the gene for X” for a small number of X.
So, the technology is just there — 23 and me is only the latest to start “pulling” on it. How will it hit consumers, medical consumers in particular?
As the new generation of tests arrives “in the clinic” something interesting happens: Most lab tests you can get are very narrow: a doctor is testing a diagnostic hypothesis and the lab test will test just that hypothesis in some narrow way. The new genetics tests, though, are very broad spectrum. A doctor might be interested in just 10 base pairs but, for the same price, the test will give 800,000 or 1,000,000 or eventually the whole genome. For each better generation of “cheap sequencing”, roughly speaking nobody ever needs to be tested more than once.
And so, making the reasonable assumption that this kind of testing will enter the clinic: who’s data is it? That’s not a simple question — I mean, who “owns” it in the utility sense: where is it stored, when is it read and by whom?, how is it protected, etc.
23 and Me (and I say this partly on the basis of some writings by investor Esther Dyson) seems centered on the idea of moving that data out of the clinic and into some private reserve belonging to the patient. It is difficult data to understand and rationally manage, so if 23andMe is to help keep the data with the patient, they also try to provide education (including, I gather, non-traditional forms like social networking).
Now, that said, I think they are causing problems:
a) Retaining Data is a Testing Lab Sin
These services are, for consumers, testing labs. You send them spit, or blood, and they send you results.
It is well known how to create privacy-protective lab testing. You need: i) private, secure, potentially anonymizing communications paths to send samples; ii) private, secure return path for results; iii) no (identifying) physical retention of results other than by the buyer. Anonymous AIDS testing is a fine example. 23andMe is almost the opposite: it’s a highly invasive lab.
b) No Quality Standards Apparent
The quality of equipment purchased by the labs doing testing for 23andme is of little interest to consumers buying the results from those labs rather than buying the equipment. There is no benchmark here by which to judge the quality of their product.
c) Medical Product Without Proof of Efficacy/Safety
23andMe offers to inform customers what their genes tell them (according to some selection of scientific research picked by 23andMe) about their bodies, including important aspects of each user/patient’s health.
So, 23andMe is playing the physician in his role as adviser, yet adopting a radically new, self-approved, untested approach. They make health claims, yet they honestly have no evidence that making these claims helps, or even doesn’t hurt their customers (or the community).
Which brings us to:
d) Illegitimate research on human subjects.
It is an experiment for a lab to retain this data. It is an experiment whether or not the results have quality. It is an experiment whether or not the medical advice offered is harmful or not.
———————
There’s an attitude I sense from the 23andMe web site. While the tech and business are all very serious, of course, the initial offering invites customers to regard them as a kind of “chemistry set” toy: Ok, this might or might not lead to real hard science but it’s a fun chance for customers with some disposable income to get to play, indirectly, with some of the new genomic sequencing tech. They could argue plausibly, I think, that at least for now they are targeting a particularly self-aware sort of customer who has, indeed, arrived to join the experiment.
Still, a less “vertical” democratization of the technology will be a much more important win. Truly anonymized testing. Democratized ways for users to manage large chunks of private, digital data. Education and analytics services that help users make protective choices about what data to export, and informed choices about what can be traded for that data.
Breaking that verticality is especially important because the “education and analytics” requires just a little bit of computing and a lot of knowledge: it is “almost non-rival”. In other words, the education, analytics, and even social organizing functions should be wildly competitive markets, ideally with many, small, community-based practitioners. Sure, sharks can get into that tank, too, but at least there’s less chance of a single point of widespread failure.
Perhaps it’s just that stunt-hacking of elites to join the low-numbers of people who have so-far published their sequence left that crowd with a little too casual an attitude about what kinds of privacy are still within technological reach?
-t
Here is a nice thing to imagine – I think this fits the vision of some of the folks behind 23andMe but I don’t think it fits their current implementation:
I go to the doctor:
Me: “Doc, I got this condition, see….”
Doc: “I’d like to run tests X, Y, and Z….”
Me: “Hold on a sec…. ” [plays with pda for a while] “I’ve got results for X and Y and could get Z but my medical concierge is suggesting you don’t actually need X and Y but could just use W instead. I’d rather not give you my X and Y results. On the other hand, my diagnostic adviser is asking if you wouldn’t mind checking my Q and R stats, while I’m here (we’re looking at the possibility of condition S, in case you have any thoughts on that).”
-t
To have one’s earthly origin and destiny reduced to a vial of spit seems like an affront to human dignity.
And I can no longer sit back and allow the international communist conspiracy to sap and impurify all of our precious bodily fluids!
[groan]
I was surprised to hear you were a bio-luddite Nick. Spitting on these remarkable advances in human understanding may delay the inevitable for, say, 15 minutes, but then it’s on to the an interesting future characterized with an increasingly robust understanding of what it means to be human. I’ll spit to that!