23andMe Is Monetizing Your DNA the Way Facebook Monetizes 'Likes' | Hacker News Reader

23andMe Is Monetizing Your DNA the Way Facebook Monetizes 'Likes'

climateerinvest.blogspot.com

155 pointschmars145 comments

So on one side journalists lament lack of new drug development & just the shoddy state of healthcare tech, while as soon as any new data that can potentially unlock new drug targets (23andMe) or provider better support (Google DeepMind Health) is introduced, Such efforts are quickly demonized with spread of Fear, Uncertainty and Doubt.

Rather than asking important questions such as at which level the "aggregate" data is shared, or what safeguards exists, the blog post and source "Gizmodo" article is quick to jump gun and spread fear (You are the product / you are being sold!!!). These posts are targeted to derive same emotional response that drive anti-vaxxers / anti-GMO nuts.

These blog posts are designed to generate clicks sadly.

The 23andMe data is amazingly useful for academic research too. I am familiar with a study (but not involved), soon to be published, that used their data to identify genetic variants responsible for a particular problem with pregnancy. The price was phenomenally lower than a more typical genome wide association study (GWAS) of which there are several publicly funded efforts. The results are much cleaner than previous studies simply due to the large number of individuals involved. The privacy controls are very stringent too.

No issue with the fact that it's a convenience sample?

They've also been offering to pharma companies the ability to help find cohorts of patients to make clinical trials much cheaper.

Say we have a drug we think works in a sliver of patients with muscular dystroph; we can either go screen tons of diagnosed patients to find the cohort we think will respond, or we can go ask 23andMe to send some patients with the required genetic profile our way.

Actually from everyone I talked too, 23andMe is pretty useless since they do very shallow sequencing, the amount of actual correlation is fairly low, and false positives and false negatives is actually quite high.

They don't do sequencing at all. What they're doing is using DNA SNP chips which will identify your genotype in the several million positions that tend to vary from person to person.

Although the variant data is sparse, the utility of the data depends on the study design and what is being studied. In the one case that I am familiar with, the data turns out to be very powerful, but there are some things about the study that may be unique.

I just want to pay for a service or product and have that be the end of my interaction with the company as well as the end of the company's use, possession, and monetization of my data. When I buy a stick of gum I deprive the store of that stick of gum in exchange for money. On the Internet I'm purchasing something but the company gets to hold on to and share it indefinitely. It not only benefits from the money exchange but also continues to use the very product they sold me to continue to beef up their profit. Taking my money is apparently just a nice bonus but not what they're really after. I wish the privacy cost was stated in as clear terms at transaction time as the monetary cost, and I believe that users should receive royalties when their data is used for any continued access.

Your genetic data is nearly worthless, even to you, unless the aggregated database exists. The real cost is in the analysis and interpretation not the collection of the data. Thus, for your own benefit (i.e., clear knowledge of your genetic risks), they must monetize your data (i.e., find someone to pay them to perform the analysis, whether drug company or charity, e.g., American Heart Association). As long as they are up front about what they are doing and protecting your identity, your interests and theirs are fairly aligned. I know a study funded by a famous non-profit (you have heard of them, I am sure) that advances the field of study substantially, but it is not yet published (and I am not directly involved).

>Your genetic data is nearly worthless, even to you, unless the aggregated database exists.

It's not about stealing from X or Y individual -- it's about privatizing and locking down an immensely useful and valuable global resource -- that "aggregated database".

> Your genetic data is nearly worthless, even to you...

Apparently Genentech disagreed 60 million times, and twelve other drug/genetics companies disagreed possibly the same number of times.

For only 800k instances of you.

for your own benefit (i.e., clear knowledge of your genetic risks)

It's not at all clear to me that knowledge of my genetic risks is beneficial to me, on balance.

Knowledge of risk is only beneficial if it means something can be done to mitigate that risk. For many diseases this is not the case. However such knowledge could certainly work against me in other ways, such as being denied employment or insurance.

And the privacy of my genetic data is not worth anything to those other parties, but it's sure worth something to me.

What makes you think that's ever existed?

Have you ever entered your phone number or member number at a business for a discount? You think they aren't tracking that? What about just tracking you by some hash of you CC info?

I'm sure there were 19th century New York shopkeepers that noticed a particular influx of specific immigrant nationalities in their shops, and may have inquired further and tailored their goods slightly.

Every good business owner has used info about their customers to improve their business. It's impossible to have a transaction without leaving some information.

This, like most "new" things enabled by technology, is a matter of scale. We need considered and appropriate limitations to protect us, not knee-jerk reactions that actually cause harm because we haven't considered everything.

Personal liberty is important, but it doesn't trump all considerations. If we can actually utilize the information to help research new treatments and further our knowledge of the human body, allowing the collection, and careful protection and anonymozation of the data is for the greater good, and worth the risk Imo.

This is a good thing to legislate though.

Do you really compare DNA and phone numbers? If you want to pay more for insurance, give some company your DNA and install a GPS tracker in your car (provided by the insurance company)... Anyways, for phone numbers, if I don't need to be contacted by a company, I give one of my free number that never rings... they can still track me but...

> > > I just want to pay for a service or product and have that be the end of my interaction with the company as well as the end of the company's use, possession, and monetization of my data.

> Do you really compare DNA and phone numbers?

I think you've let the prior context of the discussion influence your interpretation of my point. My comment had two two main points. First, that we've never really had a system where transactions did not convey some additional information about us, our preferences, and the market (what's more, "my data" is an interesting and often under-explored topic, IMO. If data wants to be free, and you go spreading it around, do you really have a singular claim to it anymore?). Second, there are competing interests of the individual and society when we are talking about aggregating data that might yield substantial benefits for people in general (as in the human race). We've made considerations about this before, such as with vaccination.

I'm not making a case that a private company should be able to keep all info about your genetic makeup and profit from it privately. I am taking a position that we should carefully consider the possible benefits of aggregating genetic info, the possible downsides in loss of privacy and other possible problems, see if there are ways we can increase the benefits and/or decrease the downsides, and try to make an informed decision, with the minimum FUD possible.

> Have you ever entered your phone number or member number at a business for a discount?

But I can choose not to.

"May I have your email for our records?"

"Thanks for asking, but you may not."

> Personal liberty is important, but it doesn't trump all considerations. If we can actually utilize the information to help research new treatments and further our knowledge of the human body, allowing the collection, and careful protection and anonymozation of the data is for the greater good, and worth the risk Imo.

That's fine, and it brings to mind organ donation, but this should be absolutely crystal clear. If I'm giving them my DNA, I want to know exactly how it can be used, how it will be anonymized, to whom they can sell it, whether I can opt out at some point, etc.

Personal liberty absolutely ought to trump considerations here; one reason I haven't sent a DNA sample to 23andMe, etc., is because I can't control what happens to my data after that point, and I'm not even sure if I can find out.

> Personal liberty absolutely ought to trump considerations here;

Not before we actually know what those considerations are. It's prudent to error on the side of caution, but erroring on the side of caution doesn't mean we ignore the issue after that. If there's serious public good that can be brought about, we should investigate that and explore ways it can be achieved acceptably.

> one reason I haven't sent a DNA sample to 23andMe, etc., is because I can't control what happens to my data after that point, and I'm not even sure if I can find out.

Yes, and I agree with this outlook. When I said it would be a good thing to legislate, I meant that we should investigate ways we can implement clear rules such that we can all benefit from the useful medical information while protecting privacy as much as possible. We already do this to some degree in the US with HIPAA laws, which facilitates the sharing of client medical information between medical professionals but also comes with a bunch of very specific safeguards and requirements.

>> > one reason I haven't sent a DNA sample to 23andMe, etc., is because I can't control what happens to my data after that point, and I'm not even sure if I can find out.

https://www.23andme.com/privacy/

Research participation

23andMe offers customers the opportunity to participate in a new way of conducting research (at home and online). Participating in our research is completely voluntary. Customers can choose not to consent to research and doing so will not impact their 23andMe experience.

> If I'm giving them my DNA, I want to know exactly how it can be used, how it will be anonymized, to whom they can sell it, whether I can opt out at some point, etc. ... one reason I haven't sent a DNA sample to 23andMe, etc., is because I can't control what happens to my data after that point, and I'm not even sure if I can find out.

https://www.23andme.com/privacy/#drawer-researchparticipatio...

Customers can opt in or opt out of our research at any time. If you opt out, we will discontinue using your information for research within 30 days.

Do you destroy this information once the owner has chosen to opt out?

The reason I, and I suspect many others, avoid these types of services is the risk of this information being inherited or possessed by an entity that would use this information in ways to which I do would not consent nor approve. This information would not only affect me personally, but also my children.

Germans are more privacy-aware than most for a good reason. I dread a future where an evil government gets a hold of this data and the technology to effectively use it.

Have you ever had more than a minor annoyance result from some company's use of your data? I understand the instinct and ick factor, but I've rarely seen an actual practical consequence.

I've asked this question before and it is typically answered with nothing more than hypothetical "nightmare" scenarios.

there are plenty of situations that only exist in nightmare scenarios. that doesn't make them any less scary. someone brought up seat belts further up, and it's a decent example. why wear a seat belt? it only protects you in nightmare scenarios

Wearing a seatbelt protects you in very real and measurable ways. Crashes happen many thousands of times per year. You can't say the same about your anonymous data being used.

Fatal car crashes are common, and car crashes that would have been fatal without a seatbelt are extremely common.

A seat belt protects you from real & practical scenarios, not just nightmare scenarios.

You wouldn't know what the practical consequences are because you are entirely in the dark where the data goes.

That is kind of the point.

By "practical consequence" I mean "something that has actually happened to you or someone you are aware of," not the worst case scenario of a dystopian vision.

If you get mysteriously denied a job, insurance, etc, you won't necessarily know. In fact, since you give out so much data about yourself so often, it's likely impossible for you to know who sold you out exactly or that you were sold out at all.

(As an aside, it is already illegal in the US to deny someone a job or insurance because of their DNA)

It's not illegal to deny life insurance, AFAIK. And in many cases that's more important than health insurance.

If I'm stricken with a terminal illness, health insurance won't do much for me. Life insurance could mean the difference between my family being OK or really struggling.

I'm not sure how that's relevant on meta data. Life insurance typically requires that you disclose up front if you have a terminal illness, and even if you do they can give you a graded life insurance policy.

FYI the Genetic Information Nondiscrimination Act of 2008 prohibits the use of genetic information in health insurance and employment.

Have you ever had more than a minor annoyance from not wearing a seatbelt?

You could say the same about, for example, the NSA wiretapping your calls and listening in on them. Until you get caught in some bizarre situation as a result, you'll be happily oblivious. But your rights are still being violated.

That data (your profile) is yours, and yours alone. No one has the right to package it and sell it without your consent. Now, it is possible that you may feel gregarious and offer up your data; but that would be by choice.

The amount of harm that a government can do to you is virtually unlimited. The amount of harm most companies can do to you is very small and effectively restricted by the law, even in the very unlikely event that they have some incentive to harm their customers.

The insurance industry's entire profit model is based on harming their customers -- underpaying claims

My impression was their profit model was collecting premiums from a very large number of customers, while only paying out a relatively small number of claims.

Adding to this comment, that's exactly why your rates go up when you get in an accident. Your "trustworthiness" has gone down and the insurance company might have to spend more on you in the long run.

So the logic is that insurance companies intentionally cause accidents for their members, in order to increase premiums and increase their costs?

That doesn't make any sense.

No. My point was that insurance company's profit model is making most of their money off people who don't have accidents and spending some of that on people who do. To recoup the loss after an accident, the insurance company will raise your rate. They're spending more on you, so they want some more from you.

Causing accidents to increase premiums would be pointless and actually cause them to lose money (premiums are almost never more than the actual cost of an accident).

Think of it this way: You have 5 people who pay $100 a month. After 3 months ($1500 profit), Alice causes an accident that costs $500. The insurance company pays for $400 of that and raises Alice's premium by $25. Now, their profit for those 3 months is $1100, but they're also now making $125 a month. That $400 insurance payout is paid for from Alice, Bob, Charlie, David, and Eve's insurance payments totaling $80/person/(3months).

Raised copays are the reason some people don't submit claims. It's a question of "Do I want to pay $300 now, or an extra $25 a month for 2 or 3 years?"

But to say that their profit model is based around causing accidents or paying as little as possible is a complete lie. Sure, insurance companies can be greedy, but their profit model is making money off people who don't have accidents. Paying lesser amounts is just an added "bonus" they get when accidents happen. It's greedy, but that's not the profit model.

Sorry, I was reading in context to tamana's comment above. I know how the industry works, I was thinking that you were trying to explain more on their point, but I see I was wrong. Thanks.

> I wish the privacy cost was stated in as clear terms at transaction time as the monetary cost

Privacy Center link directly from the home page: https://www.23andme.com/privacy/

Full Policy https://www.23andme.com/about/privacy/#Full

This information isn't hard to find--you have to first not assume it doesn't exist I guess.

* Disclosure, former 23andMe employee.

I don't feel particularly happy seeing a private profit oriented corporation having the DNA data and the results of analysis for millions of people and establishing a market dealing with this data.

In the end it's your life, your defects that could be on sale and determine your worth in a capitalist society. This is scary because you fate is determined by your birth and some algorithmus that may or may not be correct and some statistical chance that you suffer of certain diseases.

From a capitalistic "I want to minimize costs and maximize profit" perspective it's a perfect fit but the results on society are scary.

On the other hand there are clear benefits to have the possibility to prevent the harm caused by diseases or to adjust your life on informed data about possible risks.

However if you life insurance can buy that data from 23andme or your employer you are fucked and all the talk of equality before the law is moot because some DNA data from you prevents you from closing contracts due to "risk mitigation".

I don't know why anyone would belittle criticism in this direction - it's something that needs to be debated.

I thought about using 23andme but this problems stopped me from doing so. This shouldn't be a for profit but these are the times.

ADA already 'prevents' job based discrimination based on DNA.* https://www.ada.gov/

An employer buying DNA on potential job candidates is opening themselves up to a world of litigation for little real gain. Similarly, health insurance may no longer discriminate based on prior health conditions due to the dreaded 'Obama Care'.

We can debate if that's enough, or if these things should be rolled back. However, at a practical level it's currently a non issue.

PS: While most of this is not a significant disability. If people treat DNA condition X as an issue then it fall under "or a person who is perceived by others as having such an impairment. "

Laws can change. Health insurance can't discriminate now, but what happens if Obamacare was repealed, or if the ADA was amended?

An ugly mess happens. Which is why I think people would be asking themselves those questions before repealing Obamacare / breaking ADA. I think there's a limit to which politicians can pull tricks on people beyond which society's self-preservation instinct kicks in and people vehemently repeal harmful ideas. I doubt you can create a dystopian future with a single vote; it needs to be done very sneakily so that people won't notice until it's too late.

All you need is a big enough crisis to get people to support suspension of the normal checks and balances. That can happen very quickly, in the right circumstances.

Obamacare / ACA didn't set that policy, the Genetic Information Nondiscrimination Act of 2008 was passed almost unanimously by Congress and signed into law by George W Bush. It's a very popular, bipartisan policy.

ACA dated back to 1990, GINA acted as clarification in terms of genetic predisposition to developing a disease in the future (potential problems) and specifically excluded life, disability, or long-term care insurance. Which arguably made the situation worse.

Obamacare prevented insurance companies from treating active genetic disorders as preexisting conditions.

> if you life insurance can buy that data from 23andme or your employer you are fucked

https://www.23andme.com/privacy/#drawer-thirdpartysharing

23andMe will not provide any person's data (genetic or non-genetic) to an insurance company or employer.

We have been long-time supporters of legislative efforts intended to prevent genetic discrimination and to safeguard individuals' genetic privacy. In the US specifically, we were active in the development of the Genetic Information Nondiscrimination Act (GINA) enacted in 2008. GINA is federal legislation that protects Americans from discrimination in health insurance and employment decisions on the basis of genetic information. GINA does not cover life or disability insurance providers.

In addition, we have supported the California Genetic Information Nondiscrimination Act (Senate Bill No. 559), which was enacted in 2011.

I really wanted to use 23andme, but they actually prohibit anonymous usage. I know of no place that offers the service I want, which is at minimum like 23andme with anonymity. Add the old-style medical stuff and/or full sequencing for bonus points.

The ideal would be that I can have a friend pay cash in a physical store to get the kit. There is no account. The package includes an ID code that I may look up on the web site to get results, obviously browsing over Tor from borrowed wireless.

Currently, it looks like I'd have to actually buy the equipment myself and sequence my own DNA at home.

It's more about 23andMe advertising that by being a paying customer you will advance science, when what they're doing is selling your data to pharma companies.

I'm a 23andme customer and the aggregate data being used like this is exactly what I expected and is a big part of the reason I signed up.

There is a lot I don't love about Pharma companies, but when they use this data to develop or better target drugs, they are advancing science.

Wouldn't you prefer if 23andMe to offer the data to all scientists and not only pharma companies paying millions?

> Wouldn't you prefer if 23andMe to offer the data to all scientists and not only pharma companies paying millions?

Why do you assume 23andMe doesn't?

https://mediacenter.23andme.com/academic-research-collaborat...

"The 23andMe Academic Research Collaboration Program allows academic researchers to access de-identified, aggregated data from the 23andMe database."

>while as soon as any new data that can potentially unlock new drug targets

Unlock new drug targets is a low bar. Unlock new affordable and publicly available, patent unencumbered drug targets is what we want.

I'll keep mentioning it every time I see it, at risk of sounding like a broken record, but why do these types of comments labeling people as "nut jobs" always float to the top of HN posts? Look through all the comments here. There are plenty of reasonable arguments one way or the other for the sharing data. It's not a done deal with everyone on one side of the debate automatically being nut jobs.

Or people don't like their DNA being used for something beyond what they paid for? After all, the customers paid 23andme for a service.

Then they can not opt-in to research. But continuing personalized results are also what you pay them (or Promethease) for.

http://i.imgur.com/kh3bOi8.jpg

> spread of Fear, Uncertainty and Doubt

I agree with your comment but, what's with the capitalization of these words?

Its typically abbreviated as FUD.

https://en.m.wikipedia.org/wiki/Fear,_uncertainty_and_doubt

Nope. I voluntarily ticked the box on 23andme when it asked me if I was ok to have my dna shared with research institutions to help improve science. But SHARING is not SELLING. I'm NOT ok with them selling it because I think that is unethical of them to charge for something I thought they would give away freely.

"Fear, Uncertainty and Doubt"

About what?

Privacy?

The blog entry is focused on monetization, not privacy.

23 also has a deal with Pfizer.

And they have hired people, e.g., from Genentech, that will help them try some drug discovery research themselves.

But does anyone outside the company seriously believe this is going to lead to anything? They are not a therapeutics company.

They are a data broker.

Why not just submit DNA samples directly to a drug company instead of a data broker?

The only thing we can call "shoddy" here is 23andme's original business model: selling genetic test results. The question some are asking is whether they knew that when they started.

With the collected data, they now have value. There will be buyers.

Whether that value is ever passed on to patients is less certain.

This is really not surprising. Media coverage of 23andMe all the way back to its founding has said that it was offering SNP testing at-cost to build a big database which can be used for all sorts of things (although they were probably too optimistic about the usefulness of it). The real scandal is how 23andMe has published appallingly little research based on it. When I signed up with 23andMe, I was totally fine with them selling my genetic data; because I expected them to be participating in research. Instead, what we get is occasional cutesy stuff, occasional contribution of a cohort to other researchers, and a few papers a year on not particularly interesting topics like coffee consumption (some hits, naturally, since everything is heritable, but not an important topic). Contrast this to the UK Biobank which has been - with a fraction of the 23andMe sample size - releasing blockbuster paper after blockbuster paper over the past year alone!

I think it may just be taking time for scientists to start working with their pipeline, which is somewhat cumbersome simply due to their privacy controls. I predict we will start seeing more studies soon. See my other comment on this thread for details about one of those studies.

That seems highly unlikely. 23andMe has been in operation since 2006. They hit UK Biobank sample size sometime around 2010. The Biobank researchers have been managing to pump out paper after paper (check out Bioxiv!) over the past half-year even though as far as I can tell, their genetics pipeline only started yielding data back around October 2015. How long are we supposed to wait on 23andMe before concluding that they just aren't making very good use of their data from the perspective of the public interest and have focused on commercialization?

The pros of using 23andMe has vastly outweighed any cons, real or hypothetical.

Here is a pro: I was able to print out a 3 page report to give to my new doctor that described what medications I had a low / high tolerance for and what medical conditions to look out for as I grow older. My doctor was floored-- and as a result we do a few extra tests every year to keep track of a potential eye condition I am at a higher risk for.

So what is the potential cons? Perhaps my DNA data is used to help make drugs to help other people? Even if it's wasn't aggregated and my data wasn't anonymous I'm still unsure what I am supposed to be afraid of.

23andme doesn't provide this data. I'm assuming you exported your data to a 3rd party service. Could you share which one?

+1 from a 23andme customer who still thinks this was a questionable "funvestment"

They used to

23andme does provide health data in other countries like Canada.

I wonder what it would take to convince them that you are in Canada.

If you signed up for 23andMe when it was considered a medical testing lab rather than a medical device you can get a lot of extra information. I've heard you can use Genotation[1] to get something similar but I'm not sure how good a job they do on explaining things or telling you which parts come from multiple large studies and which are more speculative.

http://genotation.stanford.edu/#start

Promethease (http://promethease.com/) is probably the most well known in terms of getting health trait data

There's also Athletigen (https://athletigen.com/) which focuses more on the sports science side of things.

Except you're not supposed to clinically interpret those results. That's why they got in trouble from the FDA sometime ago. And now you're putting extra stress on the health care system and opening yourself up to false positives, which may lead to some really bad outcomes.

Insurance companies finding your dna and charging you more based on it?

Already illegal.

Yeah, that will stop them.

Laws change especially when there is a lot of money to be made.

    > I was able to print out a 3 page report to give to my
    > new doctor that described what medications I had a
    > low / high tolerance for and what medical conditions
    > to look out for as I grow older

Respectfully, I've previously paid a reasonable amount of money for things that have given me medical-looking reports back[0], but later turned out to be largely fiction.

What gives you confidence in this report?

[0] Turns out food intolerance tests aren't actual allergy tests. Who knew?

Can't think of any potential cons or just don't want to? Insurance companies would benefit from this data, all kinds of public and private institutions would like to research it, and as per tradition, advertising industry will find ways to monetize it, as it has done with all information for all of history.

>So what is the potential cons?

That these 'results' have not been verified for accuracy and are largely fictitious? That's why the FDA shut them down for providing such reports/predictions. Nobody has proved these reports are valid or useful in any way whatsoever.

http://csmonitor.com/Technology/Tech/2010/0915/How-reliable-...

>In July, a report from the US Government Accountability Office presented at a congressional hearing on genetic testing roiled the waters. The GAO found that direct-to-consumer (DTC) genetic test results were "misleading and of little or no practical use to consumers." As a test, the GAO sent identical saliva samples to four leading companies. It found that disease predictions varied widely, "indicating that identical DNA can yield contradictory results" from each company.

If you'd like to send me $100 I'd be more than willing to provide you with pages of sciencey looking reports as well.

It's fairly obvious by now that 23andme is just a collector of data, not a source of useful research or actual results.

The best bit: you pay to give them your data.

In that sense Google and Facebook are more fair, they give a free service and you pay with your data.

For their customers (the ones who give them samples) 23andme is expensive infotainment, which will worse case lead to a whole pile of un-necessary tests done by a generation of empowered hypochondriacs.

I don't see why you're drawing that conclusion.

As a customer, I'm happy with what I got:

(1) Fairly high-confidence data about known "big" inherited conditions. This was useful when my wife and I decided to have a child - there were several things we could safely rule out screening for. No surprises, but "no surprises" was what we were looking for. :)

(2) lower-confidence data for less-substantiated issues (who knew? I'm a fair bit more likely to get venous thromboembolism than average. So now I take more walks up & down the aisle when I'm on an airplane.). Is this life-changing or even particularly important? Nah. But on the other hand, it costs me nothing - and probably makes me happier - to change my behavior to wiggle around more. :)

(3) Some infotainment, primarily in the ancestry information. It was interesting. <shrugs>

(4)a copy of my SNPs that I store locally and may someday upload to another analysis service.

> Fairly high-confidence data about known "big" inherited conditions. This was useful when my wife and I decided to have a child - there were several things we could safely rule out screening for. No surprises, but "no surprises" was what we were looking for. :)

I'm sure you felt the need to get a second scan from a different company to confirm the negative results because this was so incredibly important to you.

The sarcastic tone of your response suggests that you're expecting me to be an idiot and/or a hypochondriac. I may or may not be a typical 23andme customer (and wasn't suggesting I was), but we had very specific, actionable reasons for using their services:

Of course we didn't do a second scan, that would be silly. My wife carries some recessive genes for inherited conditions common among Ashkenazi jews, and because I don't know half of my ancestry, we simply wanted to confirm that I was negative for them. They're not particularly common, hence "no surprises."

Checking the embryo for diseases such as Tay-Sachs typically require chorionic villus sampling, which comes with a 1-2% risk of causing a miscarriage. Spitting into a tube and sending it for SNP testing is cheap, easy, and low-risk. Had I been a carrier, we would have had CVS performed. The 23andme results let us safely skip having an extra procedure performed. Net win.

(Note that this is not going out on a limb: screening for Tay-Sachs is strongly recommended by ACOG when both parents are known to be carriers.)

There are screening panels designed to specifically hit only the carrier status markers we were interested in, but for me, the extra ancestral information tipped the scale in favor of the more general information.

23andme does not sell clinical diagnostic services, period.

If you are going to make really important decisions regarding your health or that of your descendants then you should really look elsewhere. You're not capable of interpreting the results and without a second scan you don't even know if the results are yours.

This is really not an area where - unless you are qualified in the field - you should 'self medicate', the risks are way too large.

> The 23andme results let us safely skip having an extra procedure performed.

I think you're putting much more faith in them than is warranted, and that's where for me the risk lies of services like these (besides the very obvious privacy issues).

> (Note that this is not going out on a limb: screening for Tay-Sachs is strongly recommended by ACOG when both parents are known to be carriers.)

Yes, it is.

You keep making statements referring to _me_ that are inaccurate, but would be correct if you scoped them to "the average consumer" or "an untrained individual." First, I've got more than enough biology training to interpret it - a bachelors + a small amount of graduate study looking at analysis of chip-based genetic and protein assays. (Not my day job. I certainly wouldn't claim to be qualified to do this for anyone else.)

Second, I certainly do have high confidence that the results are mine. That's actually one of the cool things about a larger assay: I spotted a second cousin in the ancestry components of the list, and the maternal components match exactly with what I know.

The remaining issue is simply knowing what their false positive & negative rate is on the carrier status reports. I don't know, but given that we have a pretty good prior that I wasn't likely to be a carrier, it was sufficient for our purposes. (And, in fairness to 23andme, that's one part the FDA is letting them continue to provide.)

All of the concerns you've expressed make sense in general. I absolutely wouldn't encourage anyone to do this without talking to their doctor.

So, then maybe if you're so exceptional you should state these things up front rather than to add them afterwards to bolster your position.

23andme is there for the general consumer, not for you specifically.

Of course there are exceptions to every rule, I'm well aware of that and accept that there may be some extreme edge cases for which 23andme makes sense. But you make it seem like your situation is 'normal satisfied consumer' at first and then add more and more information to move it into 'extremely exceptional satisfied consumer' territory.

> I spotted a second cousin in the ancestry components of the list, and the maternal components match exactly with what I know.

http://jacquesmattheij.com/if-you-have-nothing-to-hide

>Of course we didn't do a second scan, that would be silly... I don't know half of my ancestry, we simply wanted to confirm that I was negative for them

Oh god I hope you are not serious. These services are not clinical diagnostic tests and shouldn't be used as such for a thousand different reasons.

Genetic counseling is serious business. It should be administered by a trained professional in a medical setting. The results of any genetic tests should be interpreted by a trained professional.

See below. I'm completely serious. The population prevalence of Tay-Sachs is about 0.4%. (I don't look obviously Ashkenazi - quite the opposite - so there's little reason to assume I'm in a high-risk population). Therefore, there's a pretty strong prior that we didn't need to worry about it in the first case. Even assuming that the 23andme results have a horrible false negative rate -- let's say they miss 25% of carriers -- a negative result on a single screen (where we know it was actually my data, as Jacques noted) is enough to drive our per-child risk down to about 1 in 4000.

The false positive risks are small - follow-up genetic testing using a certified lab is easy.

Note that I was creating a deliberately absurdly high number for that false negative rate. The FN rate found, for example, for 23andme's Bloom Syndrome carrier status test is much lower than what I used above. The actual post-test carrier risk when I calculated it was under 1 in 10,000. Perfectly fine for a low-risk situation like ours.

"trained professional in a medical setting [...] trained professional."

This implies that you oppose testing, because with those expenses it's just not happening. Nobody likes to pay a 4-digit sum to look at a dozen spots on the genome, and then another test and another 4-digit sum to look at a few more (separate office visit of course), and then again...

I don't opposite genetic testing, I am in full support, especially as someone who is a carrier of a very rare genetic disorder. It is serious though it shouldn't be a DIY thing just like MRIs or any other medical tests.

I don't think you meant to imply that MRIs are a DIY thing, but that sounds lovely. Do you happen to know if it would be available/affordable anywhere?

It's my body. I've done lots of things already that are theoretically supposed to involve doctors or never be done:

I've done minor surgery with a Leatherman multi-tool knife.

I don't pay a dermatologist to pop my pimples. Really, who does that???

I clean my ears. When Q-tips are unavailable or too weak, I use other tools: paperclip, screw, scissors, screwdriver, pen, pen cap...

Being "serious" does not matter. If it did, we'd have condoms installed by doctors.

I am friends with some academic researchers that recently completed a study with 23andMe. My friends designed the experiment, 24andMe researches applied their design to the data and reported the statistical outcome. For their field, the results had the most significance seen yet for the phenotype studied. The genetic variants identified in the study will now go into further testing to understand the mechanisms by which they are acting, which will take time. Nonetheless, this will be a substantial step forward for the field.

Link to the paper?

Sorry, the paper is not yet published, which is why I have been a little vague in describing it.

It's ridiculous to be stingy with your DNA information. You're dropping that stuff all over any sofa you sit on. It's not you (any more than your twin is you) and it's not yours in any proprietorial sense, it's just data. Chill and let the nice scientists study it.

So far as I'm concerned, if they can use my information to help study medical stuff, it's entirely cool by me.

That's great, but it's your choice.

Also - that my DNA is on my sofa does not mean it's being published, sold, and used for purposes beyond my expectations.

I can somewhat understand the privacy implications, but if your DNA samples are anonymous and hold a cure to cancer or some other disease it belongs to humanity, imo.

You're forgetting the other point that the poster made, that your DNA is no more yours than your grandparents', or cousins, or any other ancestor or distant relative that shares bits of their genome with you.

By the way, we are all related, and virtually everything in you is not unique. There are maybe 50-100 point mutations out of 6 billion that are unique to you.

And anyway, what's the point of your genome unless you can compare it to others. So, share already. We're all just remixes of each other.

"You're forgetting the other point that the poster made, that your DNA is no more yours than your grandparents', or cousins, or any other ancestor or distant relative that shares bits of their genome with you."

No, my DNA is distinct from theirs, and it's 'mine'. :)

Those 50-100 points identify me, uniquely.

They identify you and all your clones and twins, who are separate people. Which you may currently have none of, but that might not continue to be so.

The data is anonymized, who cares?

Good for 23andMe. We could learn very important things from that data in aggregate. We need more aggregated data in healthcare.

Anonymization of data is not a simple task. Even with the noblest of intentions things can go wrong. See the problems with the sharing of medical data in the UK and examples from the US [1].

[1] http://arstechnica.com/tech-policy/2009/09/your-secrets-live...

What do you mean "anonymized"? It's your DNA, it basically is you. Much more than your name, email address, or whatever you consider de-anonimization.

It's not you until you tie it to your identity. I leave "DNA" all over; people cannot match it to me unless they have an existing sample that is definitely from me.

They can get close if they have your relative's DNA.

http://www.bbc.co.uk/news/uk-england-somerset-36245888

A man killed a girl. His DNA was gathered at the scene. 26 years later his daughter committed a minor crime, and had her DNA taken as part of her arrest, which led them back to him.

Anonymized doesn't always cut it.

https://en.wikipedia.org/wiki/AOL_search_data_leak

Would be better to link directly to the original article/source?

http://gizmodo.com/of-course-23andmes-business-plan-has-been...

Yeah, and that article was published last year ...

Right. But still makes it to the top ;)

When I signed up for 23andMe at least half of my motive was to help them discover new things about people's genomes including the development of new drugs. So good for them, basically.

What wrong with using aggregate dna data to help target cures for Parkinson's?

Even with the best of intentions things can go wrong and there are lots of unanswered questions in this emerging field. I'm sure the medical organizations involved are very aware of these problems, like protecting the privacy of those in the datasets. We've seen medical data de-anonymised in the past - it's not a trivial problem to solve. Likewise there are moral issues. What happens if you discover X% of people in the study have an extremely high risk of developing Parkinson's? Do you contact them? Do they have the right to know? Do you have the right to tell them? Would they even want to know? These are questions we're only starting to think about now.

> Do you have the right to tell them?

This I don't understand: what does it mean? What is the right to warn people of dangers? How do you earn such a right? How do you lose it?

Have someone else tell them if you think you haven't the right somehow.

23andme has a very clear privacy policy on their landing page. So what's this all about?

It's kind of funny that the blogger puts "those who speak do not know" at the top of one of his articles where he apparently has quite a bit to say.

A privacy policy is not informed consent. Nobody would have a problem if they asked first. The tech industry likes to act as if they have a right to use any data they get their hands on for any purpose; why wouldn't you expect a hostile reaction when people see their data being used without their explicit permission?

A common response is that nobody would participate with a proper opt-in system. That lack of participation is sending a clear message.

IIRC 23andme gives you a very specific "agree for this data to be used for medical research" consent prompt.

As a 23andme customer I guess this means that drugs that suit my DNA are therefore as a result of this study going to benefit me before anyone else. I don't mind if my DNA is used for scientific research and I don't call that a moral grey area in any way. It's data.

Good for them and for me.

I'm sad to see such poorly written blog post with a clickbait title get all these points.

I'd suggest that you instead read this well thought out (and rather complete) paper about the biomedical ethics at play within 23andMe's two-sided market business model.

(no paywall to download the paper) https://bmcmedethics.biomedcentral.com/articles/10.1186/s129...

This sort of data is exactly what I would not want to entrust to any organisation in the midst of all of the leaks and hacks we are seeing in the news. Let alone having them sell it legally to god knows who...

I remember reading about a hack/leak of, I think around 20,000 people's fingerprint data from the US government, and thinking to myself about the consequences of this biological data falling into the wrong hands. If there are any spies in that list and any country, say China, gets a hold of it, there is no way to update that spy's security, you can't change their fingerprints the same way I can think of a new password after a hack.

I might not be worried about anyone seeing my DNA profile at the moment, but maybe I will be in the future, and once the data is out there it can't be taken back or changed on my end.

My entire genome is publicly available through Harvard's Personal Genome Project. I weighed the pros and cons, and settled on helping contribute to progress was more important than my fear of the unknown future.

I do appreciate that there is a lot of public good being done with the information, my comment came off more doom-and-gloomy than I had hoped. It just seems a bit risky to me, its the sort of thing that can't really be anonymous and that gets my hackles up so to speak.

Me too! https://my.pgp-hms.org/profile/hu80855C

let me know if you find anything interesting. When I had it sequenced, the genetic counsellors said I had none of the common disease risk factors, which seemed a bit surprising.

I have a background in biology, have done extensive work with genomic data, and weighed the future risks pretty carefully. Ultimately I didn't see any real problem with posting my raw BAM files and I assume a dedicated person could identify me if they tried.

And if you want it to be available, it should be. I haven't seen anyone arguing that people who consent shouldn't be able to make their info public.

My concern is for people who don't want that info to be made public

The data has been anonymized (donor name removed), deidentified (records which could be used to derive the donor name have been removed or otherwise modified), and aggregated (grouped across multiple individuals). It's unlikely (although not impossible) to take that data and convert it to DNA profiles of individuals.

> I remember reading about a hack/leak of, I think around 20,000 people's fingerprint data from the US government

Actually it was 5.6 million.

https://en.wikipedia.org/wiki/Office_of_Personnel_Management...

Most people are missing that by only making this info available at market rates, rather than as a non-profit, it doesn't benefit the general good/public equally. It benefits incumbents asymmetrically.

Furthermore, by only collecting data from those with excess income, afflictions of the poor will remain less diagnosable.

There will be some trickle-down into poorer people, but if you think it "doesn't hurt to help people get healthier," try to see you're mainly helping wealthy people get wealthy people better. It's a slow-motion form of eugenics. We all know how well trickle-down theories work.

I have had 23andme for last 7-8 years; I've also paid for all their upgrades over the ye.ars

I've not gotten a single useful information out of it and I don't hope to in near future. It got a bunch of useless information like ear wax type, hair color ect right but overall its waste of time and money.

Privacy concerns over ppl exploiting raw data are overblown given its impossible to draw any useful conclusions from just raw data.

You can find bunch of raw 23andme exports on github.

Very presumptuous article, not that that's new. When I paid my $100 several years ago I hoped and expected that my data would be sold of as many times as possible. So much more good can come of it than bad.

Insurance companies can and do insist on health checks before they insure you, unless prevented by legislation. This is not new.

Does 23 and me actually sequence your DNA? I thought they just looked for a whole bunch of specific gene variants with known effects or associations.

Is it anonymized? I do wonder if the costs are or will one day be worth the personal gain?

If you're interested in personal DNA sequencing but don't want your genetic data shared (or you want to be able to analyze your entire genome, not just the small bit 23andMe does), check out Guardiome. This is their entire mission statement as a company.

Happy to participate. You're welcome, whoever benefits.

Is your DNA is now the "intellectual property" of a company?

The company sells rights to access sequence infromation from your DNA, i.e., their intellectual property, for up to 60 million.

Do you get any of that "up to 60 million"?

And why should you?

It's not like it's your DNA.

You sold it.

Well, actually you paid a fee to some company -- that has been threatened with being shut down by the FDA -- to take your DNA.

Who started this company? Ex-wife of Google founder.

According to Wikipedia the fee has jumped from $999 down to 99 then to 199.

So what's the service worth?

I don't know but I have a feeling it ends in 9.

Personal genomics is a great idea that has been around for a long time. Well before Google and Facebook.

But this company, applying "the Google approach" to biotech, was never a good idea.

That's only my opinion.

It's quite possible each of the 1 million or so donors was happy to donate their DNA "to science".

The issue I have with "the Google approach" is that the company is not doing much except collecting data and running cheap tests -- and of course marketing.

There's no world class research at 23andme. God only knows what they'll do with the money they get from selling people's personal information.

They are just middlemen, selling off people's personal information for a easy profit.

Very innovative.

Nobody is forcing anyone to use 23andme. If this bothers you don't use this service?

We didn't know they would do this when we signed up and gave them our DNA years ago?

Reply on news.ycombinator.com