ASHG 2010: Personal genomics

Selected abstracts from the 2010 meeting of the American Society of Human Genetics.

Genetics of infectious disease susceptibility and morphological traits in a participant driven cohort. D. A. Hinds, J. M. Macpherson, J. Tung, B. Naughton, A. Kiefer, C. Do, A. Wojcicki, J. Mountain, N. Eriksson 23andMe, Inc., Mountain View, CA.

We have developed a platform and cohort for genetics research combining self-reported phenotype information from web based surveys with genome-wide SNP genotype data. The platform enables us to pose new research questions and quickly collect responses to new surveys, with sample sizes for common traits that compare well with conventional genetic studies designed to query just a single phenotype. We have shown that the platform can successfully replicate many conventional GWAS findings, and can also generate novel association results.
Here we describe recent findings generated using this research platform, addressing the genetics of common infectious disease susceptibility as well as several common morphological traits. We report novel genome-wide significant or suggestive associations with several types of childhood and adult infections, as well as pigmentation traits and hair morphology. We find a strong association between reported mumps infection and rs485186, a synonmous variant in the FUT2 gene (P = 2e-12), as well as suggestive associations with cold sores and with shingles. We also find associations for hair color with rs1668619 near the EDNRB gene (P = 1e-8) and for hair curl with rs17558560, a non-synonymous variant in a keratin gene, KRT27 (P = 1e-9).


Direct from consumers: A survey of 1,048 customers of three direct-to-consumer personal genomic testing companies about motivations, attitudes, and responses to testing. D. Kaufman, J. Murphy Bollinger, S. Devaney, J. Scott Genetics & Public Policy Center, Johns Hopkins University, Washington, DC.

Purpose: Several companies sell personal genomic tests directly to US consumers (DTC). Speculation about the limitations, harms, and benefits of these DTC tests abound. To collect empirical data about customers’ motivations and experiences with personal genomic DTC tests, an online survey of customers of 23andMe, deCODEme, and Navigenics was conducted. Methods: Random samples of US customers who received DTC results between June 2009 and March 2010 were invited to participate. Survey topic included reasons for testing, attitudes about test results, health behaviors, and demographics. Respondents were asked to interpret two genetic test results shown using the DTC companies’ graphics and text. Results: Between January and May of 2010, 3,167 DTC customers were invited, 1,163 (37%) responded, and 1,048 participated. Most (87%) were White non-Hispanics; 24% had a BA degree and 54% had post-graduate education; 67% had incomes over $85,000. The top 3 reasons for using DTC genetic testing were satisfying curiosity (94% very or somewhat important), learning about elevated risks of diseases (91%), and learning about ancestry (90%, among those who got ancestry and health data); 77% said testing was important to improve their health. 1 in 3 were interested because a 1o relative was affected with a tested condition and 42% were interested in one or more specific conditions. A total of 58% learned something new to improve their health, while 9% felt they could not change their health risks. Only 29% had shared results with a healthcare provider or had a follow-up test; 18% intended to do so. As a result of testing, 34% said they were being more careful about their diet, 14% were exercising more, and 15% changed medication or supplement regimens. Although 88% agreed their risk report was easy to understand, 38% said the conclusions were too vague. Between 4% and 7% misinterpreted examples of company’s risk results, either answering incorrectly or saying they did not know. While 66% felt DTC personal genomic tests should be available without government oversight, 84% and 73% said it was important for an NGO like Consumer Reports, or the FTC, respectively, to monitor companies’ claims for scientific accuracy. Conclusion: Early adopters of personal genomic DTC tests, who tended to be highly educated, indicated satisfaction with their DTC services. Long term follow-up of DTC test users is needed to evaluate the impact of DTC testing on healthcare usage and behavior.


Assessing the Accuracy of Curation-Based Risk Prediction using Self-Reported Phenotypes. C. B. Do, M. Macpherson, N. Eriksson, J. Tung, B. Naughton, A. Hsu, A. Kiefer, M. Miyazawa, M. Polcari, J. Mountain, D. Hinds, A. Wojcicki 23andMe, Inc, Mountain View, CA.

Over the last several years, a number of direct-to-consumer (DTC) genetic testing companies have begun to offer personal genotyping services on the premise that single nucleotide polymorphisms (SNPs) can be used to predict complex disease risk. Typically, these methods work by combining estimates of individual SNP odds ratios and disease incidence data curated from the literature. While a number of groups have proposed techniques for theoretically estimating the predictive accuracy possible with collections of SNPs, to date, little is known about the empirical accuracy of these methods in practice. In this work, we perform empirical benchmarking of the predictive accuracy of a curation-based risk prediction algorithm for a diverse collection of 22 self-reported medical phenotypes using a population cohort of over 30,000 unrelated European individuals.
We assess each disease in terms of discrimination (i.e., the ability of the algorithm to correctly order individuals by increasing disease risk) and calibration (i.e., the ability of the algorithm to assign accurate probabilities of developing a disease within each risk group). We show that SNPs are mildly predictive for a wide variety of conditions, ranging from common cancers (e.g., breast and prostate) to autoimmune disorders (e.g., psoriasis and Crohn's disease), with typical area-under-curve (AUC) statistics around 0.60. In a comparison of our empirical accuracies with theoretical estimates derived from known SNP effect sizes and frequency estimates in Europeans, we find that the latter consistently exhibit upward bias, resulting from a combination of factors including case/control set sizes, rigor in phenotype definition, self-report phenotyping error, and the winner's curse. These data thus provide one of the first empirical assessments of curation-based risk prediction in a DTC setting, and demonstrate the value of self-reported phenotype data as a resource for performance evaluation.


Attitudes on DNA Ancestry Tests. J. Wagner Dept Anthropology, Penn State Univ, University Park, PA.

The DNA ancestry testing industry currently has 39 companies selling ~200 products and >500,000 customers. Yet the industry remains young and uncertain. No obvious industry-wide standards have been established regarding methods or communication of results. Where does one begin to evaluate the potential ethical, legal, and societal consequences of an industry that lacks a clearly defined product, clearly delineated market, or uniform commercial vision? The type and degree of the potential implications, however, must be dependent upon the attitudes on DNA ancestry tests held by those aware of the tests. Those attitudes are undoubtedly variable and changing, yet data on those attitudes are surprisingly scant. This anthropological study explored attitudes on DNA ancestry tests, including what motivates individuals to purchase the tests, how they value or interpret the scientific capabilities, and for what purposes they believe the results should be used. Questionnaires, designed and administered using Survey Monkey, were administered to general participants (N=176) recruited in State College, PA and to targeted participants (N=76) recruited through Facebook DNA ancestry-related groups. General results suggest individuals view the relationship between DNA ancestry and race in a more nuanced way than is frequently acknowledged. Test-takers are more knowledgeable about genetics generally. The main motivations for testing are education, entertainment, and curiosity. Results of this study do not support the oft-cited reasons for individuals avoiding DNA ancestry tests, namely fear of results, desire not to know, and privacy concerns. While individuals support use of DNA ancestry tests in research and support personal rights to access such information, individuals generally oppose most potential uses and are notably reluctant to permit such tests in litigation or law enforcement settings. Targeted results suggest individuals are motivated to get tested for genealogical purposes and are not motivated by legal purposes. Those who had not purchased tests avoided them because of cost and opportunity, not because of fear of results, preference not to know, and privacy concerns. These Facebook individuals were purchasing tests from multiple companies and reporting some changes in identity and behavior in light of the DNA ancestry testing, Results suggest the discussion about negative psychological effects of DNA ancestry testing may be overstated in the literature.