Remember those direct-to-consumer (DTC) genetic testing kits from Pathway Genomics that Walgreens planned to sell a few years ago? Probably not, because the FDA immediately started jumping up and down about it, and within a few days the plan was quashed.
Ah, early 2010, those were the days. Heck, GWAS studies looking for common variants were still all over the place! And WGS was still uber-cool just because we could do it in less than a year and for way less than $1 billion. The possibilities were so heady that it didn’t seem to matter much that ponying up $70-$250 for the Pathway kit would have yielded exactly nothing useful. Well, of course it did matter, but anything that smacked of personalized medicine had a certain cachet.
An odd sort of nostalgia for the Walgreens days washed over me recently as I read a Nature News piece about an analysis of a 23andMe exome sequence. Of course 23andMe is still a force in DTC genomics and has an expanding portfolio of very interesting research projects. Nonetheless, the sequence data it generates remains largely incomprehensible, more than 30 months after the Walgreens fiasco.
Gabe Rudy, a DNA analysis company executive, took a journey through his 23andMe-generated exome sequence data with a skill and thoroughness that very few can achieve. (His blog shows what I mean, if you care to go into the weeds.) What he found was interesting. Starting with 152,000 variant calls in the raw data, places where his DNA differed from that in the reference genome, he used his own expertise and specialty software, including a tool from his own company, in a step-by-step process to winnow the list of meaningful variants down to three.
Yes, three. Two are unknown and one has been classified as pathogenic, though Rudy is healthy, so its penetrance is obviously less than 100%. Rudy’s conclusion? Variant calling is “still largely a research field, and tools are built by researchers who are trying to push the limits of their data.” In other words, if you’re healthy, getting WGS or an exome sequence is still highly unlikely to reveal anything useful or even relevant, no matter how thorough the analysis.
This conclusion was immediately reinforced by a paper in American Journal of Human Genetics from December 6 that found the average healthy person generally carries 400 potentially damaging DNA variants and two known to be associated with disease traits. So mere possession of these variants is not enough to cause disease in most cases, and we don’t yet understand the tipping points for disease. The paper got a lot of media play, including on NPR, maybe because people are still holding onto the notion that genome sequences are robustly predictive for our health states. But it’s long been clear that DNA is not deterministic for health, as previously discussed here. It’s now clear that, for most of us, sequence data alone is and will likely remain insufficient to differentiate between health and disease.
And this brings up another point. These commentaries and research conclusions might make me a little nostalgic, but they in no way make me less optimistic about the current and growing potential of clinical genomics. There’s a tendency for mass media to use a sort of on/off switch, in that a negative or discouraging conclusion in one area impugns an entire field. Sequencing for disease is often useful and revealing already, and it will only become more so.
For non-patients, however, genome sequences are definitely not clear windows into our state of being. There are too many confounding factors and variables, in our environments and behaviors as well as in our own biology. So healthy people will just have to wait to see if and when benefits and understanding from WGS will come their way. But that’s OK. Healthy people have time to wait.