COMPANIES that offer analyses of future health risks based on basic genetic tests should explain the limitations of their predictions more clearly, says genomics pioneer Craig Venter.
He and four colleagues have proposed guidelines for the industry after assessing the results of scans of their personal genomes, as provided by the Californian firms 23andMe and Navigenics. They found that for each set of results, the companies identified genetic variants consistently at least 99.7 per cent of the time, but diverged on their assessment of the associated health risks (Nature, vol 461, p 724).
鈥淔irms identified genetic variants consistently, but diverged on assessments of the linked health risks鈥�
Advertisement
For four common diseases, the companies agreed for all five individuals on whether each had a reduced, average, or increased risk. But for seven other conditions, they came up with different answers for at least two of the group. The predictions were particularly varied for the skin disease psoriasis: in Venter鈥檚 case, 23andMe put his risk at more than four times a typical person鈥檚, whereas Navigenics said it was just 25 per cent above average. Such differences arise largely because they look at different collections of genetic markers.
Venter, who heads his own genetics institute in San Diego, California, says firms should agree on core markers with strong effects for each disease, and tell customers what proportion of their genetic risk these account for. 鈥淲e do have estimates, and we will consider how best to communicate them to our customers,鈥� says Andro Hsu of 23andMe.
Geneticists still know little about how differences in people鈥檚 DNA affect health, says Venter: 鈥淢y whole genome is out there and it can鈥檛 give me much more information than the personal genomics companies.鈥� What鈥檚 needed, he says, is the collection of detailed health data from thousands of people who have had their genomes fully sequenced.