Are We Close Yet?

See allHide authors and affiliations

Science  27 Apr 2007:
Vol. 316, Issue 5824, pp. 516
DOI: 10.1126/science.316.5824.516c

Large-scale genome-based surveys that look for correlations of phenotype with genotype typically examine large numbers of individuals; the results often depend on assumptions, which may not always withstand close scrutiny, about the underlying structure of the populations from which these individuals are drawn. Building on analysis of variance tests that assess whether the observed variation between populations is significant and on cluster analytic methods, Nievergelt et al. introduce the generalized analysis of molecular variance (GAMOVA). This approach extends a previous technique known as the analysis of molecular variance by creating a genetic background distance matrix and applying it to a multivariate regression analysis to test hypotheses about population structure. Several large human data sets (Centre d'Etude du Polymorphisme-Human Genome Diversity Project; Howell's craniometric characters; and HapMap) were reanalyzed with GAMOVA in order to demonstrate its potential for detecting population-level structure even among individuals in regions of low population differentiation. — LMZ

PLoS Genet. 3, e51 (2007).

Navigate This Article