The Science and Business of Genetic Ancestry TestingDocument Transcript
The Science and Business of Commercially available tests of genetic
ancestry have significant scientific limitations,
Genetic Ancestry Testing but are serious matters for many test-takers.
Deborah A. Bolnick,1* Duana Fullwiley,2 Troy Duster,3,4 Richard S. Cooper,5 Joan H. Fujimura,6
Jonathan Kahn,7 Jay S. Kaufman,8 Jonathan Marks,9 Ann Morning,3 Alondra Nelson,10 Pilar
Ossorio,11 Jenny Reardon,12 Susan M. Reverby,13 Kimberly TallBear14,15
t least two dozen companies now The Impact of “Recreational Genetics” African communities. Other Americans have
market “genetic ancestry tests” to Although genetic ancestry testing is often taken the tests in hope of obtaining Native
help consumers reconstruct their described as “recreational genetics,” many American tribal affiliation (and benefits like
family histories and determine the geo- consumers do not take these tests lightly. financial support, housing, education, health
graphic origins of their ancestors. More than Each test costs $100 to $900, and con- care, and affirmation of identity) or to chal-
460,000 people have purchased these tests sumers often have deep personal reasons lenge tribal membership decisions (7).
Downloaded from www.sciencemag.org on October 18, 2007
over the past 6 years (1), and public interest is for purchasing these products. Many indi-
still skyrocketing (1–4). Limitations
Some scientists support It is important to understand what these tests
this enterprise because can and cannot determine. Most tests fall into
it makes genetics acces- two categories. Mitochondrial DNA (mtDNA)
sible and relevant; oth- tests sequence the hypervariable region of
ers view it with indiffer- the maternally inherited mitochondrial
ence, seeing the tests genome. Y-chromosome tests analyze short
as merely “recreational.” tandem repeats and/or single nucleo-
However, both scientists tide polymorphisms (SNPs) in the
and consumers should paternally inherited Y chromo-
approach genetic ances- some. In both cases, the test-taker’s
try testing with caution haplotype (set of linked alleles) is
because (i) the tests can determined and compared with hap-
have a profound impact lotypes from other sampled individu-
on individuals and com- als. These comparisons can identify
munities, (ii) the assum- related individuals who share a com-
ptions and limitations mon maternal or paternal ancestor,
of these tests make them less informative viduals hope to identify as well as locations where the test-
than many realize, and (iii) commercializa- biological relatives, to vali- taker’s haplotype is found today.
tion has led to misleading practices that rein- date genealogical records, However, each test examines less
force misconceptions. and to fill in gaps in family histories. than 1% of the test-taker’s DNA and sheds
Others are searching for a connection to light on only one ancestor each generation
specific groups or places in Eurasia and (8). A third type of test (DNAPrint’s Ancestry-
1Department of Anthropology, University of Texas, Austin, Africa. This search for a “homeland” is ByDNA test) attempts to provide a better
TX 78712, USA. 2Departments of Anthropology and particularly poignant for many African- measure of overall ancestry by using 175
African and African-American Studies, Harvard University,
Cambridge, MA; 3Department of Sociology, New York Americans, who hope to recapture a history autosomal markers (inherited from both
University, New York, NY; 4Department of Sociology, stolen by slavery. Others seek a more parents) to estimate an individual’s “bio-
University of California, Berkeley, CA; 5Department of nuanced picture of their genetic back- geographical ancestry.”
Preventive Medicine and Epidemiology, Loyola University
Chicago Stritch School of Medicine, Maywood, IL; grounds than the black-and-white dichotomy Although companies acknowledge that
6Department of Sociology, University of Wisconsin, that dominates U.S. racial thinking. mtDNA and Y-chromosome tests provide no
Madison, WI; 7Hamline University School of Law, St. Paul, Genetic ancestry testing also has serious information about most of a test-taker’s ances-
MN; 8Department of Epidemiology, University of North
Carolina School of Public Health, Chapel Hill, NC;
consequences. Test-takers may reshape their tors, more important limitations to all three
9Department of Anthropology, University of North personal identities, and they may suffer emo- types of genetic ancestry tests are often less
Carolina, Charlotte, NC; 10Departments of Sociology and tional distress if test results are unexpected or obvious. For example, genetic ancestry testing
African American Studies, Yale University, New Haven, CT; undesired (5). Test-takers may also change can identify some of the groups and locations
11University of Wisconsin Law School, Madison, WI;
CREDIT: N. KEVITIYAGALA/SCIENCE
12Department of Sociology, University of California, Santa how they report their race or ethnicity on gov- around the world where a test-taker’s haplo-
Cruz, CA; 13Department of Women’s Studies, Wellesley ernmental forms, college or job applications, type or autosomal markers are found, but it is
College, Wellesley, MA; 14Department of American Indian and medical questionnaires (6). This could unlikely to identify all of them. Such infer-
Studies, Arizona State University, Tempe, AZ;
15Department of Environmental Science, Policy and make it more difficult to track the social expe- ences depend on the samples in a company’s
Management, and Rhetoric, University of California, riences and effects of race and racism (6). database, and even databases with 10,000 to
Berkeley, CA; USA. Genetic ancestry testing also affects broader 20,000 samples may fail to capture the full
*Author for correspondence. E-mail: deborah.bolnick@ communities: Tests have led African-Ameri- array of human genetic diversity in a particu-
mail.utexas.edu cans to visit and financially support specific lar population or region.
www.sciencemag.org SCIENCE VOL 318 19 OCTOBER 2007 399
Published by AAAS
Another problem is that questionable sci- because they exhibit different gene variants stake in them. Unfortunately, peer-review is
entific assumptions are sometimes made (11–13). Furthermore, some of the most difficult here, because most companies main-
when companies report results of a genetic “informative” AIMs involve loci that have tain proprietary databases.
ancestry test. For instance, when an allele or undergone strong selection (14), which makes As consumers realize that they have been
haplotype is most common in one popula- it unclear whether these markers indicate sold a family history that may not be accurate,
tion, companies often assume it to be diag- shared ancestry or parallel selective pressures public attitudes toward genetic research could
nostic of that population. This can be prob- (such as similar environmental exposures in change. Support for molecular and anthropo-
lematic because high genetic diversity exists different geographic regions) or both. logical genetics might decrease, and historically
within populations and gene flow occurs The problems described here are likely disadvantaged communities might increase
between populations. Very few alleles are responsible for the most paradoxical results of their distrust of the scientific establishment
therefore diagnostic of membership in a spe- this test. For instance, the AncestryByDNA (18). These tests may also come up in medical
cific population (9), but companies some- test suggests that most people from the settings: Many consumers are aware of the
times fail to mention that an allele could have Middle East, India, and the Mediterranean well-publicized association between ancestry
been inherited from a population in which it region of Europe have Native American and disease, and patients may ask doctors to
is less common. Consequently, many con- ancestry (15). Because no archaeological, take their ancestry tests into consideration when
sumers do not realize that the tests are proba- genetic, or historical evidence supports this making medical decisions. Doctors should be
bilistic and can reach incorrect conclusions. suggestion, the test probably considers some cautious when considering such results (19).
Downloaded from www.sciencemag.org on October 18, 2007
Consumers often purchase these tests to markers to be diagnostic of Native American We must weigh the risks and benefits of
learn about their race or ethnicity, but there is ancestry when, in fact, they are not. genetic ancestry testing, and as we do so, the
no clear-cut connection between an individ- Thus, these tests should not be seen as deter- scientific community must break its silence
ual’s DNA and his or her racial or ethnic affil- mining the race or ethnicity of a test-taker. They and make clear the limitations and potential
iation. Worldwide patterns of human genetic cannot pinpoint the place of origin or social dangers. Just as the American Society of
diversity are weakly correlated with racial and affiliation of even one ancestor with exact cer- Human Genetics recently published a series of
ethnic categories because both are partially tainty. Although wider sampling and techno- recommendations regarding direct-to-con-
correlated with geography (9). Current under- logical advancements may help (16), many of sumer genetic tests that make health-related
standings of race and ethnicity reflect more the tests’ problems will remain. claims (20), we encourage ASHG and other
than genetic relatedness, though, having been professional genetic and anthropological asso-
defined in particular sociohistorical contexts Effects of Commercialization ciations to develop policy statements regarding
(i.e., European and American colonialism). In Although it is important for consumers to genetic ancestry testing.
addition, social relationships and life experi- understand the limitations of genetic ancestry
References and Notes
ences have been as important as biological testing and the complex relation between 1. H. Wolinsky, EMBO Rep. 7, 1072 (2006).
ancestry in shaping individual identity and DNA, race, and identity, these complexities 2. J. Simons, Fortune 155, 39 (2007).
group membership. are not always made clear. Web sites of many 3. Thirteen/WNET New York, African American Lives,
Many genetic ancestry tests also claim to companies state that race is not genetically “Episode 2: The Promise of Freedom,” press release
(27 July 2007).
tell consumers where their ancestral lineage determined, but the tests nevertheless pro- 4. P. Harris, Observer [London], 15 July 2007, p. 22.
originated and the social group to which their mote the popular understanding that race is 5. Motherland, “A Genetic Journey” (Takeaway Media
ancestors belonged. However, present-day rooted in one’s DNA (17)—rather than being Productions, London, 2003).
6. A. Harmon, New York Times, 12 April 2006, p. A1.
patterns of residence are rarely identical to an artifact of sampling strategies, contrasting 7. B. Hoerner, Wired 13 (2005).
what existed in the past, and social groups geographical extremes, and the imposition of 8. A. Yang, Chance 20, 32-39 (2007).
have changed over time, in name and compo- qualitative boundaries on human variation. 9. K. Weiss, M. Fullerton, Evol. Anthropol. 14, 165
sition (10). Databases of present-day samples Because race has such profound social, polit-
10. C. Rotimi, Dev. World Bioethics 3, 151–158 (2003).
may therefore provide false leads. ical, and economic consequences, we should 11. S. Tishkoff et al., Nature Genet. 39, 31–40 (2006).
Finally, even though there is little evidence be wary of allowing the concept to be rede- 12. A. Mourant, A. Kopec, K. Domaniewska-Sobczak, The
that four biologically discrete groups of fined in a way that obscures its historical Distribution of the Human Blood Groups and Other
Polymorphisms (Oxford Univ. Press, London, 1976).
humans ever existed (9), the AncestryByDNA roots and disconnects it from its cultural and 13. M. Hamblin, A. Di Rienzo, Am. J. Hum. Genet. 66,
test creates the appearance of genetically dis- socioeconomic context. 1669–1679 (2002).
tinct populations by relying on “ancestry It is unlikely that companies (and the asso- 14. J. Akey et al., Genome Biol. 12, 1805–1814 (2002).
informative markers” (AIMs). AIMs are SNPs ciated scientists) deliberately choose to ancestrybydna/ethnicities.
or other markers that show relatively large (30 mislead consumers or misrepresent science. 16. M. Shriver, R. Kittles, Nature Rev. Genet. 5, 611 (2004).
to 50%) frequency differences between popula- However, market pressures can lead to con- 17. DNAPrint, Frequently asked questions, no. 1,
tion samples. The AncestryByDNA test exam- flicts of interest, and data may be interpreted www.ancestrybydna.com/welcome/faq/#q1.
18. J. Reardon, Race to the Finish: Identity and Governance
ines AIMs selected to differentiate between differently when financial incentives exist. in an Age of Genomics (Princeton Univ. Press, Princeton,
four “parental” populations (Africans, For scientists, these incentives include paid NJ, 2004).
Europeans, East Asians, and Native Ameri- consultancies, patent rights, licensing agree- 19. In contexts such as gene mapping and genomewide asso-
ciations, genetic ancestry information can protect against
cans). However, these AIMs are not found in all ments, stock options, direct stock grants, cor- confounding by population stratification or provide evi-
peoples who would be classed together as a porate board memberships, scientific advi- dence of the population origin of specific susceptibility
given “parental” population. The AIMs that sory board memberships, media attention, alleles (21). These applications are much narrower than
determination of individual ancestry.
characterize “Africans,” for example, were cho- lecture fees, and/or research support. Because
20. K. Hudson et al., Am. J. Hum. Genet. 81, 635 (2007).
sen on the basis of a sample of West Africans. scientific pronouncements carry immense 21. M. Enoch et al., J. Psychopharmacol. 20, 19 (2006).
Dark-skinned East Africans might be omitted weight in our society, claims must be carefully
from the AIMs reference panel of “Africans” evaluated when scientists have a financial 10.1126/science.1150098
400 19 OCTOBER 2007 VOL 318 SCIENCE www.sciencemag.org
Published by AAAS