[soneca]

My opinion is that this test is useless at the least and dangerous at the most. It provides information that in almost the totality of the cases no one can correctly interpretate and transform in actionable health advice. Not scientists, not doctors, much less consumers.

But it is sold as a cutting edge scientific resource that will improve your life. It wont. Not even increasing the chance that you might avoid something somehow, that's the fallacy.

For our level of knowledge regarding causality in biology and genetics, I believe this test is as good as buying your astrological map.

[Florin_Andrei]

> I believe this test is as good as buying your astrological map

Yeah, that's vastly overblown, sorry.

Let's say the average for the whole population of getting Disease X is 10% over a lifetime. You do the genetic test, and it turns out you carry an allele that's been shown with good confidence to raise that risk to 30%. What you get is not certainty, of course, but a place in a row of statistical buckets. There are studies providing solid evidence that Disease X can be typically delayed by years or decades if you do A, B, and C.

Is that information not valuable to you?

Note: I am in the general situation described above. Well, most people probably are, too, one way or another, they just don't know it.

EDIT: Perhaps you're confused because 23andme does not provide this kind of information now. But they did provide it in the past, before the FDA ban. Looks like the ban is now being gradually rescinded, one step at a time, which is good.

[DanBC]

> There are studies providing solid evidence that Disease X can be typically delayed by years or decades if you do A, B, and C.

> Is that information not valuable to you?

It depends what A, B, or C are. If it's full mastectomy based on a misunderstanding of statistics then no, that information isn't useful and might be harmful.

[kevindqc]

Full mastectomy is kind of excessive, no? I would just be more vigilant for lumps and test more.

[nommm-nommm]

Getting a double mastectomy is 90% effective in preventing breast cancer in high risk patients. Getting tested for cancer is 0% effective in preventing cancer. The best it can do is possibly increase your odds of survival.

Angelina Jolie wrote in an OP-ED about her choice to undergo a preventative double mastectomy.

https://mobile.nytimes.com/2013/05/14/opinion/my-medical-cho...

[Florin_Andrei]

I was specifically referring to the non-broken cases where there are no misunderstandings and the studies are valid.

[thinkmoore]

Oh neat, do you have a list of those? /sarcasm...

[TTPrograms]

Thanks to the 23andMe test I found out have I think 6 of 8 rare bad variants of various genes that give me what's currently estimated at around a cumulative 50% chance to develop a particular type of leukemia late in life. Of course, 23andMe doesn't provide this information - I came across it accidentally while studying my SNPs with other third party tools.

I don't really think about it that much - experimental treatments for it are already basically curing the disease and if I do get it it probably won't be for another 20 or 30 years. But now I can be the tiniest bit proactive - getting my CBC checked every few years, and watching out for unusual symptoms like fatigue.

The challenge is in communicating these things properly with fair analyses of the probability - we have X model that predicts Y risk, and we have Z error bounds on that. While it's unfortunate that most people lack the education to interpret statistical statements like that, I don't see why that should legally preclude me from getting information relevant to my health without a gatekeeper geneticist to hold my hand through "this variant has 1% prevalence, that means that 1 in a hundred people have it...".

[fudged71]

This is the purpose of Genetic Counselling. A friend of mine does this and I can certainly see the value in having someone who understands the science walk you through the outcomes.

[mrfusion]

I see genetic counselors as gatekeepers to the information about our own bodies. Just give me my results and let me figure it out for myself. If some people want the service of a counselor let them get it but don't bar people from their own bodies.

We research other issues ourselves and the world hasn't ended. Imagine if you weren't allowed to read the Bible but had visit a "bible counselor" if you wanted to find out about things in the Bible. (I use that example because people make serious life decisions based on the contents of the Bible all the time. )

[zimzam]

Not sure if you're trying to be ironic but that's exactly how it worked for many hundreds of years: remember that historically speaking, outside the Church most people couldn't read. One source estimates that in 1300 CE only 6% of England's population was literate! "Bible counselors", aka priests, were indeed the only way for most people to get information from the bible.

[cariaso]

It didn't matter that they were literate, since printing a bible in english was still illegal. It was only in 1526 that https://en.wikipedia.org/wiki/William_Tyndale created the first bible in English, allowing the common folk to go around the 'Bible counselors'. As a result the church had him convicted of heresy and executed by strangulation, after which his body was burnt at the stake.

[fudged71]

You can absolutely analyze the data yourself, a genetic counselor is a value-add that can provide an ethical medical recommendation based on the data.

[mrfusion]

My friend got a genetic test for her son and they refused to give her the results until she saw a genetic counselor.

[TTPrograms]

It's one thing to say "this has value for some people" and another to say "you're legally obligated to do it this way (costing $x more)".

[tlb]

Do you believe that no genetic tests at all can be useful? Like, even for Huntingtons or Cystic Fibrosis? Or is it just these specific tests you don't trust?

[cjbprime]

soneca has a good point here (perhaps accidentally), which is that you'll notice that 23andme doesn't return Huntington's or CF results, and it sounds like the reason is because they would be so useful and predictive of disease.

Which does put 23andme's health results in this realm soneca described of "things that might be interesting, but can't be very actionable because they don't want to scare you by returning actually actionable information to you".

[Florin_Andrei]

No. What's really happening is that they did provide a lot of probability estimates for some pretty serious stuff - back in the day before the FDA told them to stop doing it. I know because I carry a higher risk allele for a non-trivial disease (along with lower risk alleles for some other non-trivial diseases), and the test placed me in those corresponding statistical risk buckets which are different from the general population.

Then the FDA came down on them like a ton of bricks.

Now it sounds like they're reopening that door, slowly, step by step. I'm all for it.

[nommm-nommm]

>I carry a higher risk allele for a non-trivial disease (along with lower risk alleles for some other non-trivial diseases), and the test placed me in those corresponding statistical risk buckets which are different from the general population.

How do you know this is accurate?

https://mobile.nytimes.com/2013/12/31/science/i-had-my-dna-p...

>23andMe said my most elevated risks — about double the average for women of European ethnicity — were for psoriasis and rheumatoid arthritis, with my lifetime odds of getting the diseases at 20.2 percent and 8.2 percent. But according to Genetic Testing Laboratories, my lowest risks were for — you guessed it — psoriasis (2 percent) and rheumatoid arthritis (2.6 percent).

>In the case of Type 2 diabetes, inconsistencies on a semantic level masked similarities in the numbers. G.T.L. said my risk was “medium” at 10.3 percent, but 23andMe said my risk was “decreased” at 15.7 percent. In fact, both companies had calculated my odds to be roughly three-quarters of the average, but they used slightly different averages — and very different words — to interpret the numbers. In isolation, the first would have left me worried; the second, relieved.

[cjbprime]

I agree with your assessment of the situation, but I'd still be very surprised if e.g. a Huntington's test shows up in 23andme, and I think that's because the FDA doesn't trust consumers to react appropriately to receiving serious and actionable information from 23andme.

[Florin_Andrei]

> the FDA doesn't trust consumers to react appropriately to receiving serious and actionable information from 23andme

So, that is an interesting and valid point. I've thought about it myself. Seems like these tests are something new, and it may take a while before the new thing is absorbed into the culture and it's treated the way it should. Yes, there is still the risk of misunderstanding the information you're getting.

There are also a few simple solutions - e.g. any doctor should be able to help you correctly integrate this information.

[d4l3k]

23andme actually does return Cystic Fibrosis results. They just don't in the US due to FDA regulations.

I'm Canadian and have access to the full list of 23andme's health results. I actually found out that I have a recessive CF trait. Not 100% sure that its correct, but definitely something that I'm going to check out before having kids.

I'd much rather have known about that now, than after having a kid with CF.

[soneca]

I am a total layman in biology, genetics, and health, so I have no broader or deeper opinion at all on the subject. I specifically don't believe that this 23andme (or similar) test is any good for your health or useful at all. And selling it as such is dangerous.

[onewaystreet]

Then what are you basing your opinion on?

[soneca]

On articles that I read on the subject (the subject of 23andme tests, not genetics in general) and my own capacity for acquiring information and transforming them into learnings and opinions that I form to better live my life.

[searine]

This test was extremely useful to me, as I found out I was a carrier for a previously unknown genetic disease.

It was because I used 23andMe that I now know I am a carrier and can potentially prevent having a child who would suffer from this disease. I'd say that was worth 200 bucks.

[mnw21cam]

23andme can only detect common variants. In general, common variants do not cause non-mild diseases - certainly nothing worth avoiding a child being born. You should certainly not be making decisions like that based on a report from 23andme.

[searine]

It can detect whatever variants they put on the SNP chip, be they common or rare. The only caveat is we have to know what variants to put on the chip in the first place. That's the hard part, linking rare variants to disease. After that link is made, the rest is easy.

Is it a sure fire shot to detecting all rare genetic disease? Of course not.

Is it a good way to become aware that I am a carrier for a rare mendelian genetic disease? Yes. Very much yes.

[mnw21cam]

They still have to have good clustering of samples for each variant they are testing. That is, there needs to be a sufficient number of samples in each of the three states (homozygous normal, heterozygous, and homozygous variant) for the software to look at the analogue signals and draw a box around each of the three clusters. This generally prevents the analysis of rare variants using chip technology.

I would be very curious if you could look up the variant that 23andme have reported you have on http://exac.broadinstitute.org/ and let us know the "Allele Frequency" from there.

The thing is, most people are carriers of several rare mendelian diseases. When we do whole exome sequencing, we get on average a couple of hundred rare potentially pathogenic variants in each patient, of which possibly 20 or 30 are actual causes of rare mendelian disease. However, the patients are only carriers of these disease, and are completely irrelevant to their health. The likelihood of them having children with someone who is a carrier for the same condition is very slim (but something we investigate every day).

[searine]

>Allele Frequency

For the particular SNP I am heterozygous for in this particular gene, the population level allele frequency is 0.00001649. Missense mutation. Another variant in this gene, a stop loss, is at 0.0017

Yes, of course the likelihood of it being a problem is slim, but definitely not zero. High enough that it is worth knowing.

>but something we investigate every day

You and me both.

[samirillian]

Reminds me of an EconTalk podcast on epidemiology. http://www.econtalk.org/archives/2015/11/robert_aronowit.htm...

[dionidium]

This is the state of just about every trendy consumer health product. My watch tells me my heartbeat? Great. What the hell am I supposed to do with that information?

Exactly nothing.

[owlninja]

It can be pretty helpful for exercise.

[astrange]

https://www.yahoo.com/tech/apple-watch-helps-man-survive-pot...

It's also good for lessons on why not to drink too much coffee on Adderall.

Actually, this is a lot more useful than checking your own genes once they've already produced you. Generally you've already found all of that out.

[gravelc]

I agree with this to a point. Obviously, genetic information can be quite informative for non-complex traits or diseases caused by variants in only 1 or a few genes.

The reality is the genome isn't the complete instruction set for what makes you you. The interaction with the environment dataset is missing, along with any heritable epigenetic information. Add to that, our understanding of function is still extremely limited. We still don't really understand how one gene generates different proteins (via alternative splicing) at different rates. Or how gene expression is so finely regulated at a cell-specific and sub-tissue-specific level. The list goes on. Rules that seem to apply to one gene or gene family don't apply to others.

One day, we'll likely get there, but that day is still some way off. For that reason, I see no great reason to have my genome sequenced at the moment, though it'd be reasonably trivial to do so.

[mnw21cam]

> Obviously, genetic information can be quite informative for non-complex traits or diseases caused by variants in only 1 or a few genes.

And (unfortunately) those are the exact variants that services like 23andme do not detect.

23andme uses an array chip. These only detect common variants, in locations that have been pre-planned while designing the chip. A batch of patient samples are all tested together, and the probe for each variant produces a signal. Software then tries to cluster samples into three groups, which are homozygous normal, heterozygous, and homozygous abnormal. If the variant isn't common, then there would not be a decent number of samples in each group, and the clustering would fail. These tests are literally incapable of detecting any variant that is rare.

To detect rare variants, you need to do proper sequencing, for example with Sanger (single gene), or high throughput sequencing (AKA NGS, Next Generation Sequencing). This can be targeted panels of selected genes, whole exome sequencing, or whole genome sequencing.

A disease caused by a variant in a single gene (a monogenic disease) is usually caused by a rare variant. The more severe the disease, the more rare the variant is. A gene may have loads of common variants that do not cause disease. A gene may have a really rare variant that does not cause disease. Or it may have a rare variant that does cause disease - but this needs to be determined by someone with training and experience in the field.

As a lab, we regularly get inquiries by people saying "I have a variant detected by 23andme in <known disease gene> - could this be causing my <rare genetic condition>?" The answer is "No - this test is incapable of detecting disease-causing variants - it only detects the benign ones."

[astrostl]

My entire family dies of strokes, and 23&M identified me as having a blood clotting disorder. This is extremely valuable to know in terms of basic lifestyle modification (e.g. being even more sensitive to deep vein thrombosis on long flights), and in being able to discuss prior to things like surgery.

[julian_1]

People also use 23andme and similar services for reasons other than providing health advice.