Now, after nearly two years, 23andMe is announcing on Wednesday that it will begin providing customers with health information again, though much less than before and with F.D.A. approval.
The new health-related information 23andMe will provide is called carrier status. That relates to whether people have genetic mutations that could lead to a disease in their offspring, presuming the other parent has a mutation in the same gene and the child inherits both mutated genes. There will be information on 36 diseases, including cystic fibrosis, sickle cell anemia and Tay-Sachs.
I'm preparing a blog post on the differences between the old and new 23andMe reports, and how they compare to a raw data analysis with Enlis Genome Personal.
Here are the headline numbers-
Number of health-related traits reported:
Old: 201 New: 36 RawData: 2109
Number of health-related variants reported:
Old: 1283 New: 100 RawData: 13,537
I will also note that while they increased the price, they are still using the same genotyping chip that they released in December 2013
I don't know anything about this particular filing, but the FDA is primarily concerned with you having demonstrated safety and efficacy.
For this reason, I would suspect the new reports will not include anything that they could not provide scientific evidence for effective differentiation, as well as anything without convincing argument that a report would not increase workup rates, etc. with some inherent risk.
There have always been partners you can get the entire suite of results from.[1] I don't know how they haven't been restricted, but my guess is the FDA can't ban telling you what a SNP does, but the FDA can ban 23 and me selling it as a service connected somehow? Anyways, you can find out everything you wanted to know and a million other things you probably will have trouble of fully understanding.
The international version that only covers heritage still is priced 99$. The local versions outside of the US that provided health coverage already were priced higher before.
It seems to indicate a high level of naivety on the part of 23&Me in terms of the regulatory requirements around genetic testing.
What Hibbs found was that 23andMe and the FDA were not even speaking the same language. The FDA is guided by laws and regulations that limit what it can say, and 23andMe wasn’t picking up on the agency’s cues. FDA officials would leave 23andMe an opening to change the way carrier testing was regulated, for instance, and 23andMe would ask an unrelated question.
A part of me thinks it wasn't naivety and it was more "move fast and break things".
23andMe falls under Federal regulations - FDA. Uber and AirBnB operate in largely local/state regulated markets. It's harder to move fast and break things with a Federal regulator.
23andMe can do whatever they want with your data. It totally boggles my mind that someone would give a private company legal, full, and complete access to what makes you, you. We're still in the infancy of understanding what we can do with our genomes. Giving all our information over to an entity without a fiduciary responsibility to protecting your personal data is just idiotic.
We're still in the infancy of diagnosing psychological state through word choice, yet uninformed people still talk to other people over chat.
We're still in the infancy of detecting disease through microbiome analysis, yet some foolish people still meet in person without signed contracts not to collect that information.
We're still in the infancy of analyzing video to detect mental state, yet people routinely go out into public without comprehensive agreements not to record or analyze their behavior.
I understand your concern about genetic information, as it's easy to understand the impact and the potential information gleaned. That said, the cost v.s. risks v.s. benefit of the 23andme service work for me - though I would not say that they should work for anyone else. It's good to think about potential future risks, but believing too much in the malevolence and competence of people from the future is often a mistake (see: lots of cold war policies).
There's a massive difference in the order ofd magnitude of information acquired by recording people's body language vs legally handing them over your genome. That sort of apples to oranges comparison does nothing for the argument.
There is no concept of "genome security" because you leave your genome everywhere you've ever been. If someone wants your genome and is willing to spend the money to sequence it, they've got it. Let's estimate 6pg of DNA per cell and that we need 100ng of DNA for sequencing. Rounded up that means 20000 cells left behind. That means you cannot leave a single hair, fingernail clipping, feces, etc anywhere that you do not trust to destroy them.
I'd pose the question, why? From a personal privacy perspective you have a lot more to worry about from the determined actor targeting you than a data breach in which your genome (likely not even linked to any other PII if they have set up their database correctly) is just another entry in a sea of data. On the other side of the the coin, those individually worthless genomes could become valuable to the furthering of medical science if you can collect them in mass and start to look at statistical trends.
You have to multiply the expected cost of a privacy violation by the probability of that violation. A targeted attack is exceptionally rare for the vast majority of people. But in a dragnet, the probability is essentially 1.
We can have the benefits of statistical research on our private information without handing the keys to the kingdom over to malicious actors.
Research using your genetic data only occurs with your permission, and it's stripped of personally identifying information (i.e presented in aggregate) when delivered to third parties. 23andMe can't, for example, sell your genome to an insurance company.
23andMe don't have a fiduciary responsibility to protect your data, they have legal one. If you still find this overly concerning, you could just use 23andMe's genetic sequencing, which when I last checked was priced at below cost, download the data and tell 23andMe to delete all personal information.
Once you have the genetic data, there are many options for performing analytics. For example https://www.promethease.com/ which uses SNPedia and isn't under the same FDA scrutiny as 23andMe and doesn't have to self-censor.
I see it as helping myself future-proof myself. If I know I have inherent risks as I age, or worse, genetic mutations that could passed to my children, personally it's worth the possible data abuse to me.
It would be very unfortunate if this turns into Gattaca though.
This is a great idea, but the general advice you get is the same thing your doctor is telling you - lose weight, lower blood pressure, wear sunblock. For specific issues you could pass down to your kids, there are physician-ordered genetic tests. The real "killer app" right now for personal genetics has to do with pharmaceuticals - how well does my body handle Warfarin, etc. That's a great application, and something that I think is better ordered by a physician at the point where you'd need to know that information.
This is tangential to your post, but, everyone that watched Gattaca in high school and remember it as a good, thought provoking movie needs to sit down and watch it again. Gattaca is a terrible movie both from a general plot and from a scientific perspective. When the "hero" of a movie is the jackass that cons his way onto a crucial role in a space mission knowing it will kill him and leave the crew a man down, it's probably prudent to start asking questions about the other aspects. As far as we can determine the genetics for ambiguous traits like intelligence or physical ability are deeply complex and cross coupled things and it grinds my gears to have the "but Gattaca" crowd crawl out of the wood work everytime for topics like research into preventing recessive diseases like cystic fibrosis comes up.
Cam you articulate a coherent risk to an individual based on freely shared genomic information? Be specific in a way that could be used to make a decision.
Early adopters accept risks that op calls "idiotic" because they believe in potential benefits of new tech. Without those early adopters we would not have new tech.
Can we ask for a Test and once we get the Results optionally have them destroy all records except the financial transaction ones?
From their privacy policy:
>How is my privacy protected?
>We will not share your individual-level information with any third party without your explicit consent
Both "individual-level" and "third-party" still leave a LOT of room for things to be shared.
- At what level is the data allowed to be shared if not "individual-level"?
- Exactly who are third parties (is Google a third party?) and who decides that? Can it change tomorrow?
- What happens in the case they are bought out by some one? If they go bankrupt? What happens to the data?
"not 'individual-level'" sounds to me like anonymous statistical information that is not personally identifiable; is that called out anywhere in their privacy policy?
I hope this means we can finally upgrade our results to the latest chip, or that they restore some of the SNPs they dropped in the later chips that make these reports impossible to compute.
I've been waiting to unlock the Alzheimer's results for ages, but it always tells me that "upgrading to the latest platform is unavailable".
The test for Alzheimers is here: https://github.com/cslarsen/dna-traits/blob/master/py-dnatra...
(the algorithm used to be patented, by the way). By closely reading the algorithm, you can probably find out yourself, given that 23andMe still lets you browse your genome.
Actually was playing with the idea of monetizing off this project, but the laws in my country prohibit me from doing so. In fact, just using the program I made is illegal unless its my own genome.
Woah. Quite interesting! Thank you. I've been increasingly interested in bioinformatics and your project seems quite nice. I will definitely give it a try.
Thanks. It's definitely a lot of fun. And, as you can see, some things are not hard at all to detect. In fact, I was surprised how little knowledge you need to be able to flick through genome wide association studies (GWAS) and casually play around with their findings. (It's also fun to see how desperately some GWAS studies seem to try to make sense of the genome --- it's a pretty young field, I guess!)
Of course, you should be pretty careful with interpreting results as a hobbyist (as I am). But probing your genome with a parser, like the one I've made, is a great vehicle for piquing your intellectual curiosity.
I could be wrong, but I think you can still export your raw genetic data, right? If so, have you checked out Promethease? I have not used it in a while, but I imagine it will tell you most if not all of what 23andMe would have on Alzheimer's (and quite possibly more).
Man, why is it so expensive? To test my wife and I would be nearly $400. I mean sure, it might help me be able to pre-treat any known trait based disease but still.. If all they are doing is running some DNA through a database, seems like the cost should getting lower, not higher.
I believe the $99 kits were sold as a loss leader to grow their database and establish themselves as a leader in the space. The actual cost to the company was/is much higher.
There's another way to look at it: Compared to medical regulated DNA tests, such as those for breast cancer risk or other types of risk factors, which can cost in the thousands per marker, 23andme is astonishingly cheap.
While I could never bring myself to spend the $199, I was a very early adopter when it went on sale for DNA day (maybe the first DNA day sale), and the rest of my family bought the next time it went on sale for DNA day a year later. Now that medical results are back, I think it is worth either price, particularly if you have any concerns about genetic health issues being passed down to you.
In my case, my father was schizophrenic (and a couple of other relatives on the same side of the family likely suffered from undiagnosed mental illness based on what I know of their behavior and the details of their violent deaths), so I was more than a little curious about whether I had any of the genetic hints that I might suffer the same fate. I don't, and knowing that I don't have the couple of markers known to 23andme was worth at least $99.
A few years after we all signed up for 23andme, my father passed away of pancreatic cancer, which can also be a genetic condition, and so I was able to check for the risk factors that are known (though my father also lacked those markers, so it's either a marker that is unknown to 23andme, or he was not genetically predisposed to pancreatic cancer in his case). That was, again, worth at least $99, to me.
I know in either case, the ability of 23andme to predict those diseases is low, and it shouldn't replace hands-on medical care and paying attention to warnings signs, but it does provide some predictive power. And, for a while they were doing original research (polling their customers and comparing data to produce reports), which was really cool and fun to see happening. I don't think they're still doing that, maybe because of the FDA ruling.
When the medical stuff went away, I stopped recommending 23andme to friends and family, as it no longer held any interest for me. The family tree functionality is neat, I guess, but I'm not Mormon, and don't have a huge amount of interest in whether I'm related to somebody interesting 200 years ago (I'm probably not; I come from very poor white trash stock).
In short, the fact that we're arguing over whether a genetic profile is expensive at $199 or at $99 is a laughable first world problem. I'd like it if more of the world could afford it, and my signing up early was part of my encouragement for that to happen; I want it to be possible for everyone to look at their genetic predispositions and such, and take some of their medical care into their own hands with more informed choices. So, I opted in early. I would have bought eventually at $199, if they hadn't ever offered the $99 deal.
But, I too, would like to see it getting cheaper. It's been many years now, since I bought in at $99. At some point, they've gotta be thinking in terms of making it more accessible to more people. There are huge swaths of people for whom $199 is not chump change.
There's another way to look at it: Compared to medical regulated DNA tests, such as those for breast cancer risk or other types of risk factors, which can cost in the thousands per marker, 23andme is astonishingly cheap.
Many of those medical genetic tests are more expensive because they directly sequence the region of the gene associated with the disorder, providing much more data (for example, for some BRCA tests, a 'marker' consists of thousands of genomic positions, where 23andme might provide only several SNPs). This is important because a lot of disease-causing mutations are private to a single individual or a very closely-related group of individuals, and would not turn up in the population-level GWASes behind the construction of the present 23andme assay.
Absolutely true. But that has nothing to do with the original comment which was about traditional law enforcement requests for information. Given that 23andme says they haven't actually given any information to the police, it seems unlikely that cooperation with the police is why their product is approved.
In my limited understanding, this means that you can now pay twice as much as before to receive much less information than before.
There must be a lab located outside the US (and with no US footprint or marketing) that will nonetheless provide full results to anyone providing a sample.
This headline (and the one from the NYTimes) is a bit misleading – 23andMe is resuming giving health data but they’re actually giving a different kind of data. This new data is related to carrier status – genetic mutations that could lead to disease in your offspring. They still hope to gain approval to give info on health risks (what they provided up until 2013), but there is no public estimate for how long it might take. It will happen, it’s just a matter of time.
This news of their FDA approval came right after news of their latest $155mm funding round which tells us that their new investors DEFINITELY knew about this before the public. [1]
I can't imagine Fidelity Capital pouring millions into a company that got blocked by the FDA, and I was shocked to hear the news that they raised that much money until today when I found out that the FDA approved their DNA screening. I mean, they were dead as in last nail in the coffin status before that.
Not only that, recent news just broke that law enforcement now have access to members' DNA from 23andMe -- the federal government must have been appeased. [2]
As Saul Goodman says: "There's always a way to oil everyone's lock."
Probably. Unfortunately, this precludes someone who is privacy-conscious from using these products, as cool as they may seem. I can't imagine how the government would put this information to anything other than malicious use.
SNP sequencing is the ultimate fingerprint, far more individually identifying than anything else as of yet. Imagine having 900,000+ criterion to cross reference to determine identity.
What happened was that the police was using some publicly available data to cross reference with cold cases. Then when they found a familiar match, the produced a warrant to get the identity of the family member that was registered. The main issue was that the info was available, so some tech just ran the comparison. However, it feels like it would be way to easy to request the information directly with a warrant and that's it. Using something like: "We need to save the women and children, here is a list of unidentified DNA samples, please compare them to all your database."
Yes. A friend of mine is currently doing this. They used a fake name and paid with a pre-paid Amex "gift card". They had it mailed to a P.O. Box. I believe they may have used VPN and private browsing while ordering and looking at the website, and obviously created a new email account with hotmail, or whomever.
Now, obviously, you're asking them to sequence your genome, so they're getting personally identifiable information. If you have relatives who have done their 23 and me work-up, then they'll be able to tell that "John Doe" is related to Gunther Kumar, or whatever. In my friend's case, that's not likely to be an issue as they have few living relatives, most of whom are very old and would have no interest in 23 and me. YMMV.
You probably can but there is so much information in there that it might be trivial (or become trivial) to know that the name is fake based on some other information. But I guess it might be harder to tie it back to the real name.
Why don't they just setup a partner outside the US? The US based company distributes the 'device' (ie the check swab) and the sequencing, and then users can export their data to a non-US website that explains what all of the SNP's mean?
What information is lacking from the new reports? And do they provide all the raw data? If so, is it possible to do your own research based on that raw data? (I'm not arguing with you - I wish they would provide all the data possible).
Welcome back to the fold, 23&Me. Are the 23&Me health reports still saying that tongue rolling is a Mendelian trait? If you don't get details like that right, it's hard to trust the rest of your health data.
This is cool, and I'm glad its back. That being said, I got about 3 minutes joy from reading my profile again. Paid for it five years ago. Still not useful for me :/
Welcome to the world of medicine techies, where you have to comprehensively prove that your idea works or gtfo. Quite different from the Silicon Valley ethos that you're used to.
Also of note, the price has gone up from $99 (as of earlier this week) to $199.
edit: This NYTimes article has more details: http://www.nytimes.com/2015/10/21/business/23andme-will-resu...
Now, after nearly two years, 23andMe is announcing on Wednesday that it will begin providing customers with health information again, though much less than before and with F.D.A. approval.
The new health-related information 23andMe will provide is called carrier status. That relates to whether people have genetic mutations that could lead to a disease in their offspring, presuming the other parent has a mutation in the same gene and the child inherits both mutated genes. There will be information on 36 diseases, including cystic fibrosis, sickle cell anemia and Tay-Sachs.