23andMe: With Big Data comes Big Responsibility

23andMe – the home genetic testing company backed by Google – have announced they want to use their data for drug development. I think this is a great idea, but a very important time for Google to remember their “Don’t be evil” policy.

What is 23andMe?

23andMe offer mail-order personal genome sequencing: you’re sent the kit, you take a swab of DNA from inside your cheek, send it back and wait for a copy of your own blueprints to appear online. What an age to live in! The data comes in the form of a ‘SNP panel’, meaning it tests for a long list of known single nucleotide changes that are common in the general population. The unique pattern of SNPs you have inherited can show you your ancestors paths across the world, as well as identify some nasty diseases you may carry or be predisposed to.

A South-African colleague of mine took one of the tests for fun. He was pleasantly surprised to find that the family rumour that his great-great-great grandmother was black were true, and that his whole family had inherited some of her black-african SNP pattern. He took great pleasure in announcing this at a family gathering, in front of some unpleasant racist relatives. Nothing annoys bigots like scientific proof that they’re ideas are bad and they should feel bad!

By 23andMe [Public domain], via Wikimedia Commons

How can this data be used for drug development?

In the world of drug development, a very common practise is to identify ‘theraputic targets’ by sequencing DNA from a bunch of people with the same disease and looking for any common factors. Most of the time, mutations cause disease by increasing, decreasing or altering the production of a particular protein. Identifying a mutation that affects a gene that produces a protein gives researchers a target. They can then try and develop a drug that replaces the deficient protein, or inhibits an overactive or faulty one. Then, they can test the drug’s efficacy in a model organism, such as a mouse they have engineered to carry the same mutation.

So far, so science. This works some of the time, which is great, but also fails frequently, which is expensive. A lot of the problems people have with drug companies stem from the fact that drug development is an incredibly expensive and extremely high risk enterprise. This leads to a culture of ‘holding information to ransom’, and refusing to share important medical discoveries with the rest of the world. 23andMe could help put this power back into the hands of the public. Alternatively, they could become just as closed off as every other Glaxo-Smith-Kline.

Is that a good thing?

I don’t subscribe to the idea that drug companies are necessarily ‘evil’ – the nature of their industry makes it almost impossible to succeed financially without failing ethically. On average, it takes around 15 years and costs $2 billion to develop a drug and get it to market. The law then says that a company owns the patent for its new drug for five years, after which they must release it for general production. How can a business profit under these conditions? Either by charging an extortionate amount for that five years, or by adopting shady practises, such as recalling ‘faulty’ drugs every 4.9 years and re-releasing them under a different name.

If we want free information and affordable medicine, drug development should be left to governments and charities, not to businesses. Alternatively, you can look to society’s hive mind do the work for you. The general public is a huge resource with inestimable computing power. There are already several highly successful crowd-sourced research enterprises:

Galaxy Zoo has allowed the general public to help in the categorization of thousands of galaxies recorded in pictures taken by telescopes and satellites. Computers are terrible at this, but the hive-mind has taken the project forward in leaps and bounds (give it a go!)

The Yellow Card Scheme allows people to self-report any adverse reactions they had to medications. Thousands of people filling out detailed reports of things like what they ate that day and how they took their medication has allowed us to make some important discoveries that could have gone undetected for years otherwise,such as the dangers of drinking grapefruit juice while on some medications.

By Citrus_paradisi_(Grapefruit,_pink).jpg: א (Aleph) derivative work: — raeky (Citrus_paradisi_(Grapefruit,_pink).jpg) [CC BY-SA 2.5 (http://creativecommons.org/licenses/by-sa/2.5)], via Wikimedia Commons

Grapefruit contains certain compounds that alter the rate at which the body breaks down some drugs, meaning that by eating it you can accidently alter your dose in an unexpected way

I like the idea of handing science back to the people. It shouldn’t be shrouded in mystery. It’s not something for a chosen few messy-haired geniuses, it’s for people who are interested in solving problems and moving forward. 23andMe has a real opportunity to gain masses of extremely rich data and make discoveries like never before. I hope that they choose to remain clandestine with this information: the more people are allowed to contribute and become involved, the richer their dataset will become.

Self-reporting, of course, has many biases. Can you remember if your great aunt died of a stroke or an aneurysm? Do you know exactly how all your extended family are related? There are going to be problems with reliability in any dataset this big. I think the more input involved of both the public reporting and by scientific analysts, the better. Robust analysis is as big an issue as self reporting when crunching big data. It can produce a lot of red herring correlations and false effects. On top of that, these 23andMe kits won’t tell the whole story: they don’t test anything close to the whole genome, and may not even detect mutations that involve large chunks of DNA. While the project has a real, genuine chance at finding some new drug targets or previously unknown interactions, the findings need to be treated with caution. The presence or absence of a mutation is just the first step in how a disease plays out: the phenotypic effect of one little change, like the one that causes sickle cell disease, can be extraordinarily hard to predict because it depends on the interaction of a whole range of factors at genetic, cellular and environmental levels.

With big data comes big responsibility. I’m going to be following this with interest and maybe buying a kit! However, perhaps more is going to come out of the huge governmental project that’s just starting to gather momentum: The 100,000 Genomes Project.


5 thoughts on “23andMe: With Big Data comes Big Responsibility

  1. The question of the rightful owners of data continues to be debated on all fronts. In 2013 The US FDA restricted 23andMe from selling health analysis with its genome package. Do you think other countries might follow suit and will this have an impact on 23andMe’s importance as a DNA data collector?

    • Health analysis is very tricky…most diseases aren’t just black and white, so I think it’s important that actual information about your health be delivered by a medical professional. These people can help identify modifying factors to a disease that the rather reductionist approach of a SNP array might not consider, and give actual, up-to-date advise about how to treat any conditions you do have, as well as some actual counselling.
      It’s really hard to say what information they should be able to give you. On the one hand, it’s your genome and your access to the information and what you decide to do with it should be your own decision. On the other hand, there are reasons besides cost that we don’t just go ahead and sequence everyone’s entire genome to improve their medical care: it could do more harm than good. Lots of diseases can manifest very differently depending on what other traits you inherit with them. We don’t know how all these interactions work and shouldn’t imply to people that a printout saying they have inherited ‘X’ will definitely lead to disease state ‘Y’…also, telling people about predispositions for diseases they may have inherited is a bit of a doubled edged sword. One the one hand if you know you have a 10% increased risk of heart disease it may galvanise you into improving your fitness early on, which would be great for you in the long term. On the other hand, some people might react poorly and feel like the results are a death sentance.

      So, I have no idea about how much health analysis 23andMe should be allowed to provide! It’s a tricky issue. I expect other countries will follow the USA with applying restrictions to the service. It is a shame, because this will certainly impact the huge data-mining potential of these analysis kits, but perhaps we shouldn’t put that above the cost of individual healthcare.
      Perhaps they should allow extremely well-anonomised data collection for a larger portion of the genome, linked to social security numbers etc and keep that just for wide-scale analysis. Very easy to turn evil with access to data like that though…

  2. I know this has little to do with what you wrote about, but I did this before the FDA stepped in. I was lucky because I get to see the results for health, genealogy and risk factors. That isn’t the fun part though, there is also all that raw data that I can sort through for no particular reason other than curiosity all organized by chromosome and mitochondrial loop.

    I know the old saying goes a mechanics car never gets fixed, but if you want to dig through your own DNA it is an awesome investment and since you write about genetics in your downtime I think you could have a lot of fun with it.

    Plus, the privacy policy means that they have no idea who my DNA belongs to (they don’t collect any real identifying information about the person giving the sample) so if one day insurance companies can screen people who have had DNA testing, they won’t know who I am if they somehow accessed the 23andme data, but I would still have all that information for myself and my future health care.

    In any case, I’m all for some sort of genetic directory for better healthcare and medicine. Hopefully the 23andme data will be available long term to find cures for things, you and I are on the same page about medicine and drug development should be left to government and charities. Frankly, I think we just need a not for profit health and medicine system implemented worldwide.

    • It’s good that it’s well anonomised – if I lived in America I would be incredibly uncomfortable with the though that my insurance company might somehow get their hands on information like this.I absolutely agree with you about worldwide healthcare. How much better would that be? Think how much more advanced we would be as a species if we weren’t forced to hide information from one another all the time.

      I have been considering getting this done myself, but haven’t so far…I would be more excited to dig into all the backwater bits of my DNA rather than just looking at a few choice SNPs, and have been hoping for an opportunity to do my own self-sequencing at work! On the other hand, there are some health issues that run in my family and I have yet to decide whether or not I want to know if I’m likely to be affected by them. I feel like I’m in quite a good position now: I am driven to make positive health and life choices in case I’m affected; yet I’m not weighed down by the certainty that I am.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s