23andMe – DNA analytics and the future of personalized medicine

DNA analytics and the future of personalized medicine


One of the most important areas to be transformed by our ability to generate larger than life data sets, and our capacity to analyze them, is medicine. 23andMe’s ability to analyze anyone’s DNA quickly and cost effectively, paired with its success in getting 800,000 people (an impressive 90% of its customers)[1] to provide their genetic data as part of an anonymized database, foreshadows a future where personalized medicine becomes possible.

23andMe was founded in 2006 with the mission of (1) providing genetic testing and interpretation to individual consumers, and (2) through these tests, creating a vast genomic database which may be used in advancing medicine.

  1. How the test works


Three quick steps to get your report – (1) order a kit, (2) send in saliva sample, (3) logon online to view your report


  1. How 23andMe differs from other genetic testing services
  • Faster and cheaper testing (Ease of data collection)

Rather than sequencing a full genome (which costs $10,000[2]), 23andMe employs single nucleotide polymorphisms (SNP) chip technology, which sequences less than 0.1% of the genome but offers genetic results for 260 factors.[3] This enables 23andMe to provide fast results, at an affordable price – today, kits cost $199.

  • Accessing a larger population (Scale of collection)

23andMe fundamentally differs from others in the genetic testing space in terms of its underlying “Direct-to-Consumer” (DTC) approach to collecting data – it is the first and only company to receive FDA authorization to market a direct-to-consumer genetic test.[4] The vast majority of genetic tests come through a doctor, and are requested for testing one, or a very specific set, of genes. Navigenics, for example, a 23andMe competitor, makes its tests available only through healthcare professionals, whereas 23andMe launched its genetic test as a DTC product. According to founder Anne Wojcicki, “We’re DTC not because it’s easy, but because that’s how you create a revolution.”[5] A key to this ‘revolution’ is being able to aggregate genomic data from far more people, thanks to a $199 price tag and DTC marketing.

  1. Value Creation & Value Capture


23andMe creates value for consumers by (1) generating quick and affordable insights which may help them better plan their healthcare, and (2) ancestral information which may help find relatives, or be of pure entertainment value. It captures value from consumers through its $199 kit fee.

While this consumer facing value creation/value capture mechanism has generated 23andMe’s revenue so far, building its large, unprecedented genomic database has laid the foundation for 23andMe to capture value in much bigger ways – by selling data and related services, and down the line, even developing its own personalized drugs. 23andMe’s Personal Genome Service isn’t primarily intended to be a consumer device, but a “front-end for a massive information-gathering operation”,[8] where its true value creation and capture lie – analogous to the search engine’s role in Google’s ultimate value creation/capture. In February,“23andMe announced the first of ten deals with big biotech: Genentech will pay $60M to access 23andMe’s data to study Parkinson’s.” [6] The potential to uncover genetic causes of hereditary diseases through pattern matching in its massive data set could pave the way for 23andMe to develop its own billion-dollar blockbuster drugs; and in March, 23andMe announced plans to start inventing medicines itself.[7]

  1. Avoiding the big pitfalls in the data

(SNP) technology

SNP is used to sample the genome at thousands of points, rather than fully sequencing it; to then correlate genomic variants commonly found in the ‘average’ population with diseases and traits. 

  • Using genomic variants means results of these reports are not nearly as predictive as looking at family history or specific genes (such as the BRCA breast cancer gene) – geneticists call their clinical validity into question.
  • Companies doing SNP-based testing don’t define ‘average’ population risks or select sets of single nucleotide polymorphisms in the same way,  as evidenced by published research (Tables 1) on taking tests across different providers, which produced vastly different results.
Source: Variations in predicted risks in personal genome testing for common complex diseases Rachel R.J. Kalf, MSc,1 Raluca Mihaescu, MD,1 Suman Kundu, MSc,1 Peter de Knijff, PhD,2 Robert C. Green, MD, MPH,3 and A. Cecile J.W. Janssens, PhD1,4
Source: Variations in predicted risks in personal genome testing for common complex diseases
Rachel R.J. Kalf, MSc,1 Raluca Mihaescu, MD,1 Suman Kundu, MSc,1 Peter de Knijff, PhD,2 Robert C. Green, MD, MPH,3 and A. Cecile J.W. Janssens, PhD1,4
  • These tests assume genes are always ‘acting’ independently of one another – they only “consider marginal effects of each SNP and fail to accommodate epistatic effects”[9] (genes having different effects in combination than individually)

Bias in sample population

  • There may be inherent selection bias in the populations that opt to take DTC tests – (1) young, more tech savvy early adopters keen to get a glimpse of their genetic information for more recreational value, or (2) people who have a family history of a serious disease, to cite a few examples. This means the average test taker will not be representative of the average population.

Incorrect interpretation of results

  • Beyond the above-mentioned risks, there is that of consumers incorrectly interpreting their results.


[1] http://www.forbes.com/sites/matthewherper/2015/03/12/23andme-enters-the-drug-business-just-as-apple-changes-it/

[2] Source: Illumina

[3] Murphy, “Inside 23andMe.”

[4] http://mediacenter.23andme.com/blog/2015/10/21/new-23andme/

[5] http://www.fastcompany.com/3022208/fda-tells-23andme-founder-anne-wojcicki-to-stop-marketing-99-genetic-test-or-face-penalties

[6] http://www.forbes.com/sites/matthewherper/2015/01/06/surprise-with-60-million-genentech-deal-23andme-has-a-business-plan/

[7] http://www.forbes.com/sites/matthewherper/2015/03/12/23andme-enters-the-drug-business-just-as-apple-changes-it/

[8] http://www.scientificamerican.com/article/23andme-is-terrifying-but-not-for-reasons-fda/

[9] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3032061/





Variations in predicted risks in personal genome testing for common complex diseases, Rachel R.J. Kalf, MSc,1 Raluca Mihaescu, MD,1 Suman Kundu, MSc,1 Peter de Knijff, PhD,2 Robert C. Green, MD, MPH,3 and A. Cecile J.W. Janssens, PhD1,4


Stitch Fix: Mastering Personalization Through Data Science


CVS Health’s ExtraCare Program: Four Feet of Data Analytics

Student comments on 23andMe – DNA analytics and the future of personalized medicine

  1. I have always been intrigued by 23andMe as I want access to information about my own body and genetics. I believe there is immense power in said informative and how it can transform my own choices about healthcare – both preventative and reactionary. Utilizing data about me, I can better understand if I am predisposed to certain things that my extended family did not know about or perhaps did not speak about. I can also draw family linkages that have long been forgone as more and more people sign up and submit their own salvia to 23andMe, which adds to the direct and indirect network effects of healthcare data. My only concern of this is how the healthcare industry may adversely utilize my information to preclude me from future insurance policies.

  2. I love the value proposition of this company: provide personal information to their customers while also trying to advance overall healthcare by utilizing their customers data. I understand trying to predict or understand what the future may hold health-wise makes some people very uncomfortable but I think this the future of healthcare. 23andme has a real opportunity to lead the way in this market, I just hope they continue to use their customer’s data in ways that help the larger population as well.

  3. Great post on a fascinating company! I think the direct to consumer strategy is a great way to collect a large sampling of data and can be empowering for individuals. I do have 2 concerns about this strategy: 1) Does the $199 really make this service that accessible for all? I think this price tag could deter a lot of people from actually taking the test. If there is such a huge potential to make money off of this data later on, why doesn’t the company give away their testing kit for free? 2) Some people might not understand the implications of getting predictive results on their health, especially when the results are negative. How can 23andme mitigate the risk of causing serious psychological harm or other adverse side effects from negative results?

  4. Thanks for the posts! A couple of quick questions
    1. Is the information created owned by 23AndMe or the customer? I am not sure that everyone would be comfortable or willing to share their personal information the way 23AndMe is sharing it with BIotech companies for further research
    2. We see in our daily lives how insurance companies are moving towards supporting patients through preventive care. For example, by signing up for a gym we get a discount on our healthcare insurtance premiums. Would an insurance company find this product attractive in order to further customize and prevent more expensive treatments for vulnerable population?

Leave a comment