Note to users. If you're seeing this message, it means that your browser cannot find this page's style/presentation instructions -- or possibly that you are using a browser that does not support current Web standards. Find out more about why this message is appearing, and what you can do to make your experience of our site the best it can be.
Science Signaling - Call For Papers

Site Tools

  • AAAS
  • Subscribe
  • Feedback

Site Search

Search Advanced

Science 20 June 2003:
Vol. 300. no. 5627, p. 1877
DOI: 10.1126/science.1083411
Prev | Table of Contents | Next

Technical Comments

Comment on "Genetic Structure of Human Populations"

Rosenberg et al. (1) described the genetic structure of 52 human populations from five continents studied at 377 short tandem repeat (STR) loci. This high-resolution study demonstrated that, using multilocus information only, the individuals of these populations could be subdivided into five stable genetic clusters that correspond to five major geographic regions: sub-Saharan Africa, the Americas, Oceania, EastAsia, and Eurasia (Europe, the Middle East, Central and South Asia). The study also partitioned the total genetic variance into components based upon differences between individuals within populations, between populations within regions, and between these five regions [table 1 in (1)]. Surprisingly, Rosenberg et al. (1) found very small differences between regions— only 4 to 5.7% of total diversity depending on regional sampling intensity, which is roughly half that of previous estimates inferred from molecular markers (2–4) (Table 1). Although the authors attributed that odd result to differences in sampling coverage among studies, we note that they did not base their analysis on the specific stepwise mutation model prevailing at STR loci, which is characterized by recurrent mutations. Ignoring the possibility that the same allelic type found in different individuals or populations may be derived from different evolutionary processes is known to lead to biased estimates of genetic structure (5–7).


Table 1. Past and present apportionments of human nuclear diversity.
Markers (ref.) No. of loci No. of population samples No. of regions Variability estimates (%) and 95% confidence intervals (12)
Between individuals within populations Between populations within regions Between regions
RFLPs (2) 79 11 5 84.5 3.9 11.7
STR (2) 30 14 5 84.5 5.5 10.0
Alu insertions (3) 21 32 5 82.9 8.2 8.9
STR (1) 377 52 5 93.2 (92.9, 93.5) 2.5 (2.4, 2.6) 4.3 (4.0, 4.7)
STR (1) 377 14 5 89.8 (89.3, 90.2) 5.0 (4.8, 5.3) 5.2 (4.7, 5.7)
STR, this study 377 52 5 87.6 (86.4, 88.9) 3.1 (3.0, 3.2) 9.2 (8.1, 10.4)
STR, this study
377
14
5
83.4 (81.2, 85.4)
5.1 (4.6, 5.7)
11.5 (10.0, 13.1)

By extracting information on the minimum number of mutations separating the alleles at all 377 loci (8), we reestimated (9) components of genetic variance under the same hierarchical structure as was used by Rosenberg et al. Our results (Table 1) differ significantly from those in (1) but are in perfect agreement with previous results obtained from other molecular markers with comparable sampling designs [with as much as 11.5% (95% CI: 10.0 to 13.1%) of total variability due to differences between regions, and only 83.4% (95% CI: 81.2 to 85.4%) due to differences between individuals within populations].

A correct assessment of variability at the local or continental level is important not only for understanding the history of human settlement, but also for designing sound strategies to discover segments of our genome undergoing selection (10). Our analysis is in agreement with a previous study of minisatellite diversity showing that reduced variance components are obtained in the presence of high levels of homoplastic mutations (11). Our analysis also shows that this bias can be removed by using the right mutation model for STR loci, and suggests that the comparison of STR allele frequencies among different regions may be misleading. The demonstration that this huge data set is not flawed underscores its importance for estimating other important parameters of human population history.

Laurent Excoffier
Grant Hamilton
Computational and Molecular Population
Genetics Lab
Zoological Institute
University of Bern
Batzerstrasse 6
3012 Bern, Switzerland
E-mail: laurent.excoffier{at}zoo.unibe.ch


References and Notes

  • 1. N. A. Rosenberg et al., Science 298, 2381 (2002).[Abstract/Free Full Text]
  • 2. G. Barbujani, A. Magagni, E. Minch, L. Cavalli-Sforza, Proc. Natl. Acad. Sci. U.S.A. 94, 4516 (1997).[Abstract/Free Full Text]
  • 3. C. Romualdi et al., Genome. Res. 12, 602 (2002).[Abstract/Free Full Text]
  • 4. L. B. Jorde et al., Am. J. Hum. Genet. 66, 979 (2000). [CrossRef] [ISI] [Medline]
  • 5. Y. Michalakis, L. Excoffier, Genetics 142, 1061 (1996).[Abstract]
  • 6. M. Slatkin, Genetics 139, 457 (1995). [ISI] [Medline]
  • 7. F. Rousset, Genetics 142, 1357 (1996).[Abstract]
  • 8. The difference in the number of repeats between two alleles was inferred by taking into account the length of the motif of each STR, and the size of the amplified PCR fragments. STR alleles with imperfect motifs were discarded from the analysis.
  • 9. The analysis of genetic structure was performed according to the model described in (5).
  • 10. C. Schlotterer, Curr. Opin. Genet. Dev. 12, 683 (2002). [CrossRef] [ISI] [Medline]
  • 11. J. Flint et al., Hum. Genet. 105, 567 (1999). [CrossRef] [ISI] [Medline]
  • 12. Confidence intervals were computed from 20,000 bootstraps of the 377 loci.
  • 13. This work was supported by Swiss NSF grant 31-054059.98 (L.E.).
Received for publication 12 February 2003. Accepted for publication 11 April 2003.


THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES:
Spectrum: joint bayesian inference of population structure and recombination events.
K.-A. Sohn and E. P. Xing (2007)
Bioinformatics 23, i479-i489
   Abstract »    Full Text »    PDF »


ADVERTISEMENT
Click Me!

ADVERTISEMENT
Click Me!

To Advertise     Find Products

ADVERTISEMENT

Featured Jobs

Science. ISSN 0036-8075 (print), 1095-9203 (online)