UC Berkeley plan to test freshmen DNA criticized | Hacker News Reader

UC Berkeley plan to test freshmen DNA criticized

google.com

15 pointsSerene4 comments

> Critics, however, worry that students could get the idea the school approves of widely available direct-to-consumer gene-testing kits that claim to predict the risk of future health problems

Presuming it's accurate, I really don't see the problem with it. It's (in theory) just a more sophisticated, accurate method of predication than we've had in the past.

I can't imagine any health risks to ring the alarm bells about either- I haven't heard of any breaking research that shows taking saliva samples or cheek swabs to be carcinogenic for example.

I especially like that Berkeley's claimed goal is simply to get students talking about it.

I'm familiar with the fears people have- designer babies, genetic discrimination, health insurance- but it seems really silly to advocate the medical equivalent of sticking our fingers in our ears and going 'lalalala I can't hear you'

The problem is not that the kits cannot accurately read off your DNA; they do an admirable job of that. The problem is that our knowledge of the impact of any particular allele is exceedingly incomplete.

Without a clinical trial validating these tools, most of us in the field find it to be inappropriate to advise someone as to their risk profile on the basis of direct-to-public kits. (I'm not talking about the BRCA tests or other validated, gene-specific tests - I'm talking about 23andMe-style genome-wide kits.)

Particularly for complex (polygenic) traits, these throw you into a minefield. Consider two modes of error: (1) The test indicates impending illness. (2) The test indicates no impending illness / good longevity. For (1), imagine that you have a disease-predisposing allele that the kit catches, which is why it thinks you'll be sick. But it doesn't catch (since these kits only target 1/10,000th of the genome) some other allele that protects you from the same disease. So you inappropriately worry about the disease. If disease risk is modifiable by behavior (such as exercise), perhaps it encourages you to work out more often. But, perhaps it will make you feel like you cannot overcome your genetic destiny, so you don't work out at all. The prophecy would then be self-fulfilling.

For (2), imagine that the testing kit sees that you have a few protective alleles for disease X - but you also have a novel mutation that is private to your family and not found in the general population. The testing company has no idea that this exists, and it obviously won't be on these testing kits (which overwhelmingly track common, not rare, variants). So now you end up getting disease X a few years down the road, despite all evidence from the kit suggesting that you'd never get it.

These are my two favorite failure modes, though there are plenty of others. At the end of the day, although I'm in the field of human genetics, I'm not going to be advising people to get these kits until there are clinical trials validating specific risk predictions. (I'm not saying not to use the kits - but I am saying that it's going to be more for fun than for accurate medical prediction. For now.)

When I said 'accurate' I meant the conclusions they draw based on your DNA sequence. From what you are saying, it sounds like (in this sense of the word 'accurate') they are not, so my point is invalid :)

The good thing is that, with each passing day, my point becomes less valid, and yours more so. The bad thing is that it will probably be a decade or more until your original point is more valid than mine. I'm in this field and would love for it to have the accuracy that many people think it already does! :-)

Reply on news.ycombinator.com