My 23andMe DNA Results | Hacker News Reader

My 23andMe DNA Results

techcrunch.com

67 pointsaneesh19 comments

Yes, $1000 is a lot of money for that test. No, it is not incredibly useful. We're in a period of time when DNA testing is mainly for early adopters. However, the Moore's Law equivalent for DNA sequencing blows away the Moore's Law for transistors: every year genetic sequencing has become cheaper by tenfold.

23andme's disease risk indicators aren't overly meaningful, because they only test for a few SNPs and true disease risk is made up of many SNPs and multi nucleotide mutations and deletions, as well as levels of miRNA expression. Many of the mutations are only studied in caucasians. For now, most genetic tests are interesting rather than explicitly useful. Five or seven years from now, after Google's 100,000 genome project is complete and you can get a whole genome test for $1000, genetic testing will be very useful. Let's just hope that the results of that study are completely public.

Is 23andMe's test for HIV resistance the same gene as your startup tests for?

That would make sense, though I thought they only tested for single nucleotide polymorphisms and the HIV mutation in question is a 32 base pair deletion.

Social software with a 'how closely I'm related to you' feature would be compelling. Social networks obviously apply, but how about when you exchange emails with someone... wouldn't it be cool if it mentioned a common ancestor in the footer? Oh, and on a dating site, automatically filter out all first and second cousins!

maybe incestdating.com ?

The limitation of SNP techniques is that the magnitude of association between a particular allele and a disease have turned out to be rather low - in the order of 1-2 times increased risk. This is virtually meaningless on an individual level - for example if there is a 1:1000 risk of a particular disease in the population and you have the risk allele you would still only have a 1:500 risk of getting the condition.

A possible explanation for this is that that all of the alleles that are tested in genome wide association studies are found relatively frequently in the population - it may be that rare alleles exist that confer far greater risk e.g. 10-100 increase and it is these that mediate the majority of the inherited risk of a condition. These rare alleles could not be detected by a microarray based approach but would be identified by massive resequencing. A preliminary resequencing effort is already underway: http://www.1000genomes.org/ and within the next few years whole genome resequencing on an individual basis will become feasible with massively paralell techniques such as Selexa.

About 10 years ago, I considered the possibility of a gadget that connected to a home computer and allowed you to perform genetic testing at home - without revealing information to third parties. They'd be a large market for such a device. Unfortunately, it isn't feasible yet but it will be.

What are the inhibiting cost factors?

For SNP testing, the cost of micro-arrays and micro-array reading machines are way too much for at home use. Whole genome testing right now is astronomically more expensive than SNP testing.

However, if the rapidly decreasing cost of genetic sequencing keeps up, then it will eventually be possible to create at home genome sequencers that feed into your home computer to analyze the data. The privacy paranoid will have to pay a significant premium for this privilege.

Also, a DNA extraction machine would have to be bundled with the at home sequencer, ideally in one machine, and DNA extraction machines are also too expensive for home use. Without a machine, DNA extraction involves proprietary solution combined with various steps of heating and centrifuging.

A fluidics station is not cheap, and neither is the rest of the typical Affy rig (and Illumina is worse), but having set up more than a few, I have to tell you, the cost of the components isn't actually very great. (As in, the parts and labor. The price paid to the company is most assuredly huge.)

If there were actually demand for it, someone could throw together the equivalent of an open-source array reader kit for (guesing, but not without basis in fact) $10K or so. God knows it's faster and easier to work around Affy or Illumina's junkware with R than it is to get anything done with their binaries... but that would be silly because who outside of a hospital or lab really has that much use for one?

But sequencing-by-synthesis methods that can use WGA on small samples are the key to a truly accessible and useful map of an individual's genome. I don't really care how my genome differs from Craig Venter's; I want to know how it differs from itself as I get older, I want to know how it differs from itself as I am exposed to environmental factors, and I want to know how it differs from the baseline for my 'eigen-ancestry' in terms of disease risk.

There are so many statistical piles of shit that have already been stepped in (particularly WRT linkage analysis, but I'm sure we'll eventually see a good laffer from a GWAS of some sort now that they're all the rage) that I'd really rather have the tools to work on my own, thankyouverymuch...

The reagents are extremely expensive and the computational requirements of the assembly require access to a cluster.

is it even possible to build something that would allow a non-technical person to conduct genetic testing at home?

Definitely. Already, some high end scientific equipment is almost completely automated. I think it could get as simple as "spit in tube, place tube in machine, push button on machine, plug machine into USB port on computer." There will exist a market opportunity for this eventually

cool

This opens up a whole new level of narcissism. "My DNA is better than yours." :)

And discrimination.

I find this kind of thing fascinating, but I'm surprised no one has commented on the health insurance angle here. I wonder how long it will be before "discount" health insurance is available if you agree to take this test. All kinds of fun implications once we go down that road...

Oh man, I had no idea something like this existed! Shades of Gattica, but SO SO cool!

Also reminds me of Dawkin's writing on a "Moore's Law" for genetics, basically saying that there's an exponential equation describing number of genes sequenced for a fixed cost.

Guess we've gotten at least to the point where $1000 buys one (apparently) meaningful information about your genetic structure.

Methinks I'll wait until the cost has fallen by another order of magnitude or so . . .

I think this might be a good chance for family members of people who suffer from rare and un-curable diseases to see if they too are susceptible. Other than that I think 100$ might have been a good price for this vanity.

Reply on news.ycombinator.com