Hacker News Reader

Comment by krrrh | Hacker News Reader

I really liked the old 23andme and was an early customer. I felt it did a good job of presenting scientific conjecture with notes on confidence and how to use the data. It gave me some encouragement to cut down on my caffeine consumption, and to exercise more to try to influence specific gene expression, and if I was a carrier for a serious genetic disease, I definitely would want access to that information.

I logged in recently after not looking for a few years and was so disappointed with what the site has become--essentially ancestry astrology with the defaults presenting a preposterous level of precision on your racial mix. I'd like to think that this was just a result of inflexible regulation from the FDA, but I suspect that they are also bending to market demand for meaningless information like how 1 or 2 of your 64 great-great-great-great-grandparents happened to share some genetics with a bunch of people from the Asian steppes.

On the other hand... I do enjoy reading about identitarians of all stripes having emotional crises when they learn that they can't claim purebred status.

View on HN

> I really liked the old 23andme and was an early customer. I felt it did a good job of presenting scientific conjecture with notes on confidence and how to use the data. It gave me some encouragement to cut down on my caffeine consumption, and to exercise more to try to influence specific gene expression, and if I was a carrier for a serious genetic disease, I definitely would want access to that information.

So I wonder what exactly in the 23andme data made you do these things. I don't know about your caffeine intake, but if you drink a lot then I could've told you to limit it without any genetic data. And I certainly could've told you to exercise more.

Don't get me wrong: I'm not opposed to genetic testing. If people want to know their genes, ok. (I have done a 23andme test myself.)

But realistically there isn't a lot of data that is useful for lifestyle changes. The idea is "If I have disease X risk in my genes I can do something against it". However "something" usually means "exercise, eat healthy, don't smoke, avoid injuries, do all the boring stuff everybody knows". You don't need a genetic test to tell you to go to the gym for your health.

Sure. Everyone knows they should exercise more, but what tips the balance in general, and what tips it on a day when you feel like crap and would rather surf Hacker News or go back to bed? Friends or family you exercise with, a loved one getting sick, a realization that your body doesn't work as it used to when you throw your back out, a gym opens up near your house? Which straw strengthens the camel's resolve?

I wouldn't say that 23andme data was the only or even main factor, but for someone with a scientific bent, having read a few studies that showed an x% improvement in odds on avoiding a disease that I didn't want in populations with specific genes that I shared provides an extra bit of motivation on some days. I already exercised, but progress in health isn't binary, it's about frequency and repeated effort. Miss a significant percentage less workouts over a few years and the $99 on genetic testing easily pays for itself.

There have been people with schizophrenia on both sides of my family; my dad was schizophrenic. I really wanted to know if I have any genes that indicate an increased risk of schizophrenia.

I signed up for 23andme on the first DNA Day sale and signed my parents up on the second or third one. They didn't have a lot about schizophrenia in their health data early on but added a few markers later. I don't have any of them (but my dad didn't either, so there's probably something missing from the picture). When my dad passed away a few years ago from pancreatic cancer, I wanted to know about that, too. Early detection is the only thing that can help with pancreatic cancer. Again, no markers for increased risk (and again, my dad didn't have any significant risk markers either, the science is still very new).

But, for me, part of the value lies in the original science they're doing. Even if they don't have useful answers for me today, I would do the questionnaires and such in hopes that it'd help somebody else later. Genetic prediction of diseases can be hugely valuable.

Sure, there's a lot of things that the standard advice works for. But, schizophrenia? Sometimes, by the time it is diagnosable the person is difficult to reach with treatment. Maybe knowing it's potentially looming will make them more likely to seek treatment at the first signs of trouble, rather than after the point where it is difficult to determine what's real. Likewise, pancreatic cancer...there are tests for it (not good ones, but if they know what they're looking for, they might catch it early), but they're not part of a normal checkup. But, if you know you've got family members who had it and have the genetic markers for it, maybe you pay for, and deal with the invasiveness, of the extra tests every year or two, in hopes that if you do get it you'll catch it early enough to have a hope of surviving more than a year.

"I really wanted to know if I have any genes that indicate an increased risk of schizophrenia."

So does everyone else. Unfortunately, the best science out there is insufficient to give you an answer. If 23andMe were doing that, they would have deserved to have their wings clipped. Merely reporting the markers would be irresponsible -- there are hundreds! Interpreting those markers is, at best, an unclear process.

The GGP comment is ironic, because the problems they are complaining about (weirdly precise ancestry claims) are exactly what happened with much of what 23andMe used to report -- exaggeration of weak statistical signals. Ancestry is something more understandable to laypeople than relative risk factors for rare diseases, so folks can more easily understand the absurdity of the claims.

Sure, it's early. I know it and knew it when I signed up. I was not shocked to find out that the markers they'd identified as being associated with schizophrenia aren't a complete picture or a way to detect schizophrenia. And, I also know that genes aren't the only thing that determines whether one will suffer from schizophrenia or pancreatic cancer or most other diseases (as far as we know). I think they always presented the information without being misleading (though the government disagreed).

But, I also know that genetic markers for some diseases are a real thing and they are useful predictors of disease in some cases. And, the number of markers that have predictive applications are increasing all the time, as research gets done.

I don't expect 23andme to replace a doctor or a psychologist. But, I took a little bit of comfort in seeing that the markers that are thought to be predictive (even if mildly so) aren't present in my genes. I don't think they ever presented them as being a way to diagnose any condition. It was always just a fun thing you can do with your own genetic data, and maybe help move research in the field forward by answering questionnaires now and then.

I enjoyed playing with 23andme when it was focused on health-related data, less so now that it's focused on ancestry. I haven't logged in for well over a year.

I mean, I was amused to debunk my family's native American myth (which I think many American families have, for some reason), and I was also able to rule out my theory that the broad noses and dark skin that run in my family came from black folks somewhere in the family tree (that surprised me, I actually expected we had some African lineage, despite no one ever mentioning it...I assumed that's why some old folks in the family claimed native ancestry, to explain away those features). So, I guess the ancestry thing is reasonably amusing, too, but I probably wouldn't have signed up just for that, since I've always known I'm an American mutt of poor white trash lineage.

"I was not shocked to find out that the markers they'd identified as being associated with schizophrenia aren't a complete picture or a way to detect schizophrenia. And, I also know that genes aren't the only thing that determines whether one will suffer from schizophrenia or pancreatic cancer or most other diseases (as far as we know). I think they always presented the information without being misleading (though the government disagreed)."

They were completely misleading (at least, the reports I saw were misleading). Most of the stuff they reported was taken wildly out of context -- if it weren't, a lay audience couldn't possibly understand it!

But even ignoring the fact that you don't know the field (and therefore cannot possibly evaluate their claims, aside from general skepticism), a big part of the reason that you shouldn't be doing your own genetic analysis is that a lot of science is crap. Even "good" science is crap. So it's not that the markers "aren't a complete picture" -- it's that they're probably just noise, and you can't possibly know, unless the test passes all of the hurdles to get certified. You need the system, because the system is the only thing that validates results over time. Strength through redundancy.

Also, for whatever it's worth: be very careful with correlating anything to their ancestry calculations. Those are pretty much crap, too.

The APOe4 (https://www.snpedia.com/index.php/APOE) gene's connection to Alzheimer's is one of the most well-studied. Roughly, if I recall correctly, one allele doubles your risk and both multiplies it even further. There's some evidence (probably weak, but still) that exercise limits that risk https://well.blogs.nytimes.com/2014/07/02/can-exercise-reduc...

Now, you could say "everyone should exercise", and maybe so. But any tangible evidence for the long-term benefits helps. Sure, there's general evidence of benefits, but in my (limited, dated) exposure to the literature, general exercise research is not necessarily compelling enough, especially given the difficulties in pulling out correlations between exercisers and other associated activities or traits.

I find exercise boring, and I'm definitely looking to minimize exercise necessary to maximize its benefits.

By the way, the genetic research on predisposition to side effects to drugs and such is also extremely compelling. It's sad that the medical world is still so far from even trying to do precision medicine.

I run a nonprofit project that gives more of the advanced data back to 23andMe users. I try my best to avoid the ethical/legal grey areas by not giving any kind of disease predictions, and instead provide a search service that links your genome to academic papers. These citations and references are what I miss the most from 23andMe's old reports.

Would love feedback! https://www.infino.me/genetics

It looks interesting. Would be great if you had the option to upload raw 23andme data rather than just connecting to the account, as some people probably have concerns with giving up their identity along with the information.

I haven't considered that would be an issue. Great point! I plan to eventually build a parser for Ancestry.com, which lacks an API but has a similar raw format to 23andMe. When that gets done I'll lump them together.

^ Agreed with above -- when I've discussed this with my more privacy-minded friends they always raise a concern about their data getting into the hands of insurance companies down the road, and having their premiums affected. Data security would be something I'd think quite a bit about, but I like the sound of what you're offering.

These issues certainly keep me awake at night. I'm really glad that we have laws like GINA that protect us from having our medical insurance change due to genetic testing. I hope that congress preserves these protections, and further applies them to other forms of insurance (like life insurance).

This nonprofit of mine was born out of a lifetime of dealing with a rather serious medical issue. With full genome sequencing, I was able to finally diagnose myself as having a rather severe mitochondrial disorder. I've been worried that knowledge of the mutation would affect my medical care, but I finally reminded myself that the mutation was already making me quite ill to begin with! The biochemical concerns are now much more worrisome than the hypothetical insurance or discrimination concerns. I've grown relieved in the realization that finally knowing whats up will likely do more good than bad.

A big problem with your site is it links to stuff as if its significant/important/real. Nearly every single variant I followed through on was basically bogus. An intronic variant with like 50% of the population having it.... Literally doesn't matter how many GWAS studies have associated that with a greater risk of breast cancer. The risk difference by definition is nothing. Its meaningless to state clearly on your site "breast cancer". You might want to avoid ethical/legal grey areas. But who is your site aimed at? It has none of the information a scientist would want, or a clinician would want... But does have misleading absolutist linking of variants with papers and diseases?

My site is aimed at people who want to dig deep into the literature. I don't say any of it is significant or real. I say "here's a paper that thinks it's real. Go read it". I can certainly do better to qualify things more clearly.

I've been working in genetics for some time and what I have noticed is that all of these markers start as an unjustified signal in GWAS studies with no biological proof. Eventually however, some clever biologists eventually figures out the role of that gene in model organisms, and the locus becomes more confidently verified. Many of the markers for obesity, for example, have been thoroughly vetted now as being significant and having a well explained role in physiology. They show up in the brain centers that regulate metabolism.

My mission for that project is to provide links/searches into the literature for educational reasons, so that I can inspire people to pursue careers in science and figure out what the unknown genes actually do.

Just because a single SNP variant has a high penetrance in the population and a low overall risk doesn't mean that the gene itself isn't extremely important. Genotype arrays like 23andMe only test a small fraction of the genome. What we are finding is that many of these common variants are co-inherited with rare variants that are the actual sites of functional importance for a trait. So just because a SNP on a genotype array has a low risk, the gene can still be profoundly relevant.

> if I was a carrier for a serious genetic disease, I definitely would want access to that information

I'd much rather get that information from an MD who is at least bound by HIPPA regulations, than by a VC funded internet startup.

Is anyone able to providing a notable source on 23andme and their position on HIPAA regulations?

Do you have a specific question? Their position is that they follow the law where it applies to them.

23andMe isn't a HIPAA covered entity, and HIPAA regulations don't apply to then at all.

I would much rather get it done anonymously through a VC funded startup than through an MD who is going to unambiguously tie it to my person, log it into an "approved" and "secure" storage and may be required to report it based on some currently unknown set of conditions ("do you have a predisposition to X? let us alert the right folks so they can check you out, just for your own benefit")

I understand that with personal genetics you are unlikely to stay anonymous against a determined effort, but ATM I personally trust my government "trying to do good" less than I trust a random stranger who wants to make a predictable profit for his business. Just my 2c.

So a VC funded startup is not "a random stranger", but doctors are the government?

Doctors are not government, but they are bound by a variety of government regulations (e.g., parent referenced HIPAA). I do trust my own physician, but he still has to report on me in certain cases or he will lose his license.

For big, uncertain queries I initiate I would rather find the answer privately and then decide how public do I want this info to be. Otherwise I am afraid that my options will be curtailed: "did both of your parents have heart problems? sorry, we cannot take you on that climbing trip", etc.

I was under the impression that 23andMe doesn't allow you to anonymously submit a genetic profile.

Some people submit their DNA but switch up their gender so as to theoretically nullify the results for a potential future data breach. (Insurance providers being one concern here.)

Does anyone know if/how 23andme addresses this, or do they care?

That... seems like literally the easiest thing to correct based on a genetic sample.

Switching the gender would be an ill-advised way to be anonymous. Its easily determined from the raw data itself, based on the presence of Y-chromosome SNPs. I have heard that some people prefer to use fake names, emails, prepaid credit cards, and use a PO box or some other mailing relay to mask their home address.

I'm not sure if 23andMe cares about anonymous users, but they might care about survey results being faked in a large number of users, because that would affect their scientific mission of finding interesting patterns.

They do indeed care.

My brother and I submitted a sample from my dying mother, which they were unable to process properly. We were offered a replacement kit, but at that point our mother had already passed away. When I conveyed this to 23andme, they refunded our kit cost and removed our mother's pending profile from my 23andme account.

From their email:

"Unfortunately it is not possible for us to unregister a kit after an initial sample fails analysis. Replacement kits are automatically registered to the individual the original kit was registered to, and they retain personal registration information including name, birthdate, and legal consent to our Terms of Service and research document.

If we allow someone else to use the replacement kit, the sample would not be associated with their personal information. Additionally that person would not agree to our Terms of Service and provide legal consent for us to process their sample. For privacy and legal reasons, a replacement kit cannot be used by another individual."

I think this is a different circumstance however. It sounds like the issue here was primarily based on obtaining proper consent from the individual being tested, which isn't the same as anonymity.

Check this out though. It seems as if their official stance is they do want your real name, but permit pseudonyms when conversing online. Interesting.

https://customercare.23andme.com/hc/en-us/articles/202907890...

> It seems as if their official stance is they do want your real name

Its because they're selling the data to pharma companies. They have a vested interest in the data being accurate.

https://www.google.com/search?q=23andme+sell+data

If you have to submit a payment you aren't anonymous.

I completely agree with your distrust of a random startup, but I encourage you to exercise more caution about your MD. Not that he may have bad intentions but because most of them aren't really computer geeks and your personal information is quite likely to end up on a lost USB stick or out in the wild because of a bad shared password.

I grew up in a family of doctors, and I heard things like them considering to sue google because a patient managed to find his medical record by googling his name. They are scientists, very educated people, but that doesn't mean they understand all technologies.

> but I suspect that they are also bending to market demand for meaningless information like how 1 or 2 of your 64 great-great-great-great-grandparents happened to share some genetics with a bunch of people from the Asian steppes.

What if some people really want to know this? My family is from the Caribbean, I will never be able to trace back my family history as deep as others because my ancestors were slaves, 23andme seems like a great service. Especially in a time with so much racial animosity in the US, maybe if people had a deeper look at where they come from they would have sympathy for others.

That's surprisingly a thoughtful observation that I never would have considered since I don't have that problem. I feel kind of bad for disregarding my carribean girlfriend's results now.

We're talking about $200,000,000 here. There's no space in the room for sentimental value with a round like that. It's got to be brutally practical and be able to scale.

oh of course. If I understand correctly, the FDA has already approved them to provide genetic testing for 10 diseases. That is the big sell they are going for. They have plenty of data from the sentimental folks like me to play with.

Yup, I agree. My mom was adopted so I can't trace anything on one side of my family. That's why I did the ancestry test. She met her biological parents before they died, but has little info about family history.

Reminds me of this (very much NSFW) Chapelle skit: http://www.dailymotion.com/video/x3a3f0f

So many of the identitarians believe in eugenics, and then find out much of their current racial makeup comes from Africa: https://www.theatlantic.com/science/archive/2017/08/white-na...

I can't say I mind the schadenfreude. Bigots of all types should be educated.

What does that even mean? Everybody's ancestors come from Africa.

TL;DNR: It means white racists got really upset when they found out this fact that they were part African per their DNA. I laughed a lot when this made the news.

As the article pointed out, 23andMe's original business model of selling genetic testing kit to consumers didn't fly after FDA's mandate of banning them and they only gain 10 related gene testing approval since 2015. So they pivoted. As the article pointed out again, I still think they have some very interesting ways to generating more revenues like making drugs(a far stretch) or gene related disease analysis (fertility, breast cancer, etc.)

What in your genetic makeup told you to cut down on caffeine? I'm a passionate coffee drinker, and have been sipping coffee the entire day since my teen years, so I'm interested in what you identified as a risk.

Perhaps https://www.snpedia.com/index.php/Rs762551

There's an allele that leads to slower caffeine metabolism.

There is a SNP that determines whether you are a slow or fast metabolizer of caffeine. Depending on your genes coffee consumption has been linked to breast cancer and heart attack rates. Mostly I always had assumed I was a fast metabolizer and could get away with drinking loads of coffee, and realized that I was pinning my metabolism and getting poor quality sleep. I was better off titrating down and finding alternatives.

https://www.snpedia.com/index.php/Rs762551

Even though 23andme doesn't present this information anymore, as noted elsewhere on this thread, there are other websites such as promethease, genetic genie, found my fitness, SNPedia, etc, which you can upload your raw data to. Some focus on specific topics.

You can always export your genetic info from 23andMe and plug it into SNPedia. The user interface isn't as nice but it has more SNPs listed than 23andMe did.

You can also export to Found My Fitness for some free detailed reports focused around diet/cardio health. https://www.foundmyfitness.com/

Agree, I expected more from them by now. I mean, I know they had trouble with FDA and it's fun to learn how many Neanderthal genes I have, and there's a bunch of interesting info they delivered, but for 6+ years I am on their site, I expected more. Maybe more useful info beyond entertainment. E.g. they say I am predisposed to weigh above average. OK, why don't you give me some useful personalized advice? What they have is "eat well and exercise". Duh, I could read it anywhere. Maybe indeed it's a consequence of FDA meddling?

I'm also disappointed. I can only recommend downloading your raw data and purchasing a report from Promethease. All the data is there, you just need another service to provide it.

This might be useful for people yearning for the good old days.

http://slatestarcodex.com/2014/11/12/how-to-use-23andme-irre...

Reply on news.ycombinator.com