Think twice before analyzing/releasing your genetic data

Think twice before analyzing/releasing your genetic data(jacquesmattheij.com)

82 points by eapen 14 years ago | 44 comments

This is not an issue in the US. The Genetic Information Nondiscrimination Act makes it illegal for insurance or employers to discriminate based on genetic information. See Wikipedia for more: http://en.wikipedia.org/wiki/Genetic_Information_Nondiscrimi...

Considering how easy it is to get ahold of someone else's DNA and the potential advantages of genetic testing, I think it makes more sense to push for anti-discrimination laws instead of avoiding genetic testing.

slowpoke 14 years ago | |

Except those laws won't do anything to prevent abuse. It's the same as with tracking data, really: the only way to prevent abuse is if there is no data to be abused in the first place. You're severely deluded if you think laws will prevent the abuse of this data, much less discrimination based on it.

jerf 14 years ago | | |

But this is a problem society is going to have to deal with head on. Trying to protect your DNA is futile. Maybe you shouldn't post the results, but it doesn't exactly take a lot of serious interaction with the medical system before they're drawing blood for something or other. And in the very near future, they'll probably want to start sequencing your genes almost as often, as we enter the era of being able to tell from your genes whether a given drug will be effective or not, etc.

I don't deny there's a problem, but this is at best a several-year stopgap, not a solution. You will want them to sequence your DNA to determine which cancer drugs will work best when you are sitting in an office with cancer.

The problem is that medical insurance is currently uncomfortably between actual "insurance", coverage for unknown and expensive catastrophic events, and "health care", a program for spreading the costs of health care between many people to equalize the burdens and also sort of cover the insurance use cases. Expanding our knowledge of what you might get starts breaking down the premises the insurance system is based on.

read_wharf 14 years ago | | |

Yes, exactly, and we know that insurance companies will find ways to harass if not abuse data when it's in their interest. Remember the health insurance companies that give employees bonuses to find ways to invalidate a customer's policy when that customer suddenly has an expensive claim.

borism 14 years ago | | |

how are you planning to get rid of your DNA?

CaptainZapp 14 years ago | |

So: what about if you apply for insurance and are asked along the lines "if you have any knowledge of a high probability of specific illness due to genetic testing?"

Sure, you can lie. But god help you if you ever need coverage [even for a non-related disease] and the insurance just flat out won't cover you.

I for one rather don't know and don't. One of the major reasons is exactly the one described above.

danssig 14 years ago | | |

Personally, I would recommend moving to a country where you don't have to worry about such fears. Any non-US first world country will do.

apaprocki 14 years ago |

There are different ways to help advance science without publishing all of your genetic material. As someone who has published my own mtDNA and my father's mtDNA to the NIH GenBank, I feel like I am meeting science in the middle. The mtDNA alone is not particularly useful for identification, but it does help population geneticists who are at a severe shortage of information for certain groups. My father's mtDNA happens to be in a rare haplogroup that only has 4 public samples.

toomuchtodo 14 years ago | |

How can I go about publishing my own genetic data to the NIH GenBank sequence database? If I've already done genotyping with 23andme, is the raw data acceptable for submission?

apaprocki 14 years ago | | |

Sadly, 23andme data will not be enough. If you want to submit mtDNA, you will need the full sequence. I used FTDNA to fully sequence the mtDNA. They give you a link to download the FASTA file after it is done and that file can be prepped to submit after you have it.

I used these instructions: http://www.ianlogan.co.uk/Submission.htm

jacquesm 14 years ago | |

That's an excellent approach.

kolev 14 years ago |

There are more positive sides than negative about sharing genome information. For reference, check out: http://www.personalgenomes.org/ - smarter than us people have already donated their fully-sequenced DNAs (23andMe, Navigenics, deCODEme, and others don't sequence your full genome for now). Paranoia never leads to progress! Anyway, there's a law that you cannot be discriminated by insurance companies and employers with your DNA. Also, with ObamaCare in 2013 you can't be turned down for insurance either. And medicine's only chance for significant breakthroughs is through genomics. And, last, but not least: my DNA is mine and nobody else's, so, if my children and grandchildren have an issue with me sharing my own genome, sorry, but enjoy what you've inherited and quit whining, bastards!

mjwalshe 14 years ago | |

Do you think that will stop the insurance industry finding ways around this.

And if you have some marker that means that your children cannot get good jobs with health insurance - congratulations you have just condemned your kids to live on the bread line.

kolev 14 years ago | | |

Paranoia can take you only that far. HN is supposed to be a place where smart people with big data in there minds hang out. Imagine when insurance companies has everybody's DNA - you think there's a perfect DNA? No, everybody has a different sets of risks. Again, just because you have a risk, it doesn't mean that you will developer the disease. For example, according to 23andMe, I have a lowered risk of Melanoma, yet, 8 years ago, I had stage 0 Melanoma. Genetically, my children are safe as their DNA is clear, but this won't prevent insurance companies tracing my kids health records to mine, if you know what I mean. If you look at the current research, risks are +/- a few percent, so, it doesn't really give you much except certain conditions. But it is much better for you personally to know what you're highly predisposed to than to be a paranoid chicken and not get your DNA sequenced.

eapen 14 years ago |

This is very helpful advice. I had considered publishing my DNA from 23andme after seeing others had posted their genetic data. I am really glad, I didn't and think its quite important that you consider others (including your siblings + kids) before publishing such data.

Update: Thanks po. I hadn't really considered the risk of submitting data to 23andme in the first place. Updated the title to reflect this as well.

OT: Despite knowing this, I am sure I would have tried the service anyway. However, after seeing the data, I haven't found it too helpful. Probably, because most of their data is based on older Europeans and I don't fall in that category.

po 14 years ago | |

I think jacques is taking it one step further and suggesting that you think twice before even submitting your data to a company like 23andme at all. It's an interesting point. I'm trying to think of what kind of privacy policy would cover all possible nefarious uses of that data. I think it basically would have to be one like "We won't give your data to anyone except you, even in aggregate and in the case we go out of business or are bought, we will destroy all of the data."

Gatsky 14 years ago |

This is important advice. There are doctors who train for many years to counsel people and guide them through the decision making process about genetic testing for diseases. However the blog post obfuscates the fact that the kind of genome sequencing available to the public online is pretty harmless, for two reasons:

A. There is no way to predict accurately individual risk of disease based on genetic information EXCEPT for a few rare diseases associated with highly penetrant mutations such as Huntington's disase, BRCA mutations (breast and ovarian cancer syndrome). Even then, taking 23andme as an example, they do not sequence enough of the BRCA gene to provide meaningful results.

B. Detailed risk profiles do not appear important to health insurance companies. You can work this out by the questions they ask you - they are mainly interested in pre-existing conditions. Using blood pressure, cholesterol and family history it is possible to generate a pretty good estimation of the risk of heart disease, but AFAIK an insurance company doesn't care about this.

It is also unlikely that in the future genomic information as it exists today will ever be able to predict risk accurately.

So don't worry. Or even better, don't waste your money and don't worry.

nodata 14 years ago |

Oddly enough, this is similar to publishing social data.

If I upload my address book to Google, they can see who I know. Often who I know is also who knows me. Without the people in my address book's consent, I and my friends have collectively published the information for people who have opted out.

darklajid 14 years ago | |

I disagree. That comparison feels wrong to me.

One is highly personal and immutable.

The other is less personal, however hard you try, and shouldn't be on the same level as rants and bilge drinking pictures.

hinathan 14 years ago |

Fascinating point about parents — knowing their genome means you could in principle know the entire potential configuration space of all their (real or hypothetical) offsprings' genomes.

dirtyaura 14 years ago | |

Even if we ignore crossovers, potential configuration space for genotype is very very large, 2^46. Add crossovers and mutations and variations in genotype is practically infinite. Of course how that maps to variation in phenotype is a complex issue.

wouterinho 14 years ago |

I had the same thought when considering sending DNA to 23andme, which is a US company. Having your processed DNA in some database exposes you to a lot of potential risks with little upside.

dawgr 14 years ago | |

Is it not possible to ask them to delete all my DNA data? My subscription will end soon, can I just ask them to delete it completely in a clear worded email to make sure there is no misunderstanding as to what I mean by "deleted"? Will EU laws protect me from this or are they irrelevant since they are a US company?

polymatter 14 years ago | | |

If you do business in the EU, you must comply to EU laws. And to remove doubt, a person physically located in EU counts as being in the EU. At least that was what I was advised, but IANAL, this is not financial advice etc.

cyanbane 14 years ago |

Naive question, but how much of this information could be gleamed by a nefarious source from an aluminum can you throw away or a piece of hair you drop?

gwern 14 years ago | |

Enough to convict you in a court of law - cops do it all the time. (It's not a search requiring a warrant if you throw the cup away, it turns out.)

cjbprime 14 years ago | |

All of it.

read_wharf 14 years ago |

As this information becomes more widespread across populations and more detailed per individual, it will degrade the current relationship between insurance and customers that makes insurance work for both sides.

This must change the way insurance companies decide to cover, and how much they charge.

Or health care itself must change from largely insurance provided, to "something else."

jmount 14 years ago |

This is an issue (even in the US and Europe, where laws are going to at best protect you from some of the possible abuses). I wrote on this also from a similar point of view: http://www.win-vector.com/blog/2008/03/do-not-let-your-medic...

MediaSquirrel 14 years ago |

Q for the OP: Does this advice apply to consumer genetic testing services like 23 & Me?

The reason I ask is that I'm considering using them to analyze my and my newborn son's genetic code. I have not combed thru their TOS with a fine toothed comb, however. Pls advise. Thx!

sycren 14 years ago |

Perhaps though by analysing your genetic data, you may also help those family members if you find a gene that suggests the likelihood of getting cancer or some other disease which, when treated early offers a better quality of life.

jacquesm 14 years ago | |

This is possible, but consider that a likelihood is definitely not a certainty and that you are at risk of contracting far more diseases with a genetic element than the ones that you will actually contract.

Carrying a gene does not automatically translate into getting a disease or even a significantly elevated chance of getting a disease. That only works if the chance is approaching near certainty, absent any symptoms if you don't use a high enough cut-off you'll be engaging in a high-tech variation of medical students disease.

sycren 14 years ago | | |

I do realise that as a bioinformatician, but we need more genetic data from multiple people to see if there are any trends with certain genes that do predict illnesses. I really would like in the future to see blood tests test for everything rather than something specific. While it will be argued that this would create a lot of false positives I think that utilising machine learning and datasets from a whole population we will start finding the probabilities of certain illnesses and will hopefully lower the chance of a false positive.

Recently a University in the UK was doing some brain scans for psychological research using volunteers. Two of these volunteers were found to have brain tumors that they would not have found out about until probably too late which is why I wish testing was more streamlined..

blake8086 14 years ago |

I don't understand how we can really expect to conceal our genetic data. Everyone leaves it everywhere all the time (lost hairs, skin cells, saliva on discarded beverage containers).

onemoreact 14 years ago | |

You could say the same things about fingerprints but it's uncommon for a company to collect your fingerprints without you knowing.

blake8086 14 years ago | | |

I think that's a good point, but the stakes are quite a bit higher with your genome, since it can reveal a bunch of information about medical costs you're likely to incur, and how much shorter or longer than average you'll live.

Joakal 14 years ago |

How would one poison pill genetic data to the point of it being worthless to associate with the individual?

cjbprime 14 years ago | |

It's probably impossible; groups have already tried to anonymize genetic data and failed, and our ability to match genotype to phenotype is only getting better.

nefasti 14 years ago |

I'll go grab my tinfoil hat and be right back.