An Arabidopsis Example of Association Mapping in Structured Samples

Abstract

A potentially serious disadvantage of association mapping is the fact that marker-trait associations may arise from confounding population structure as well as from linkage to causative polymorphisms. Using genome-wide marker data, we have previously demonstrated that the problem can be severe in a global sample of 95 Arabidopsis thaliana accessions, and that established methods for controlling for population structure are generally insufficient. Here, we use the same sample together with a number of flowering-related phenotypes and data-perturbation simulations to evaluate a wider range of methods for controlling for population structure. We find that, in terms of reducing the falsepositive rate while maintaining statistical power, a recently introduced mixed-model approach that takes genomewide differences in relatedness into account via estimated pairwise kinship coefficients generally performs best. By combining the association results with results from linkage mapping in F2 crosses, we identify one previously known true positive and several promising new associations, but also demonstrate the existence of both false positives and false negatives. Our results illustrate the potential of genome-wide association scans as a tool for dissecting the genetics of natural variation, while at the same time highlighting the pitfalls. The importance of study design is clear; our study is severely under-powered both in terms of sample size and marker density. Our results also provide a striking demonstration of confounding by population structure. While statistical methods can be used to ameliorate this problem, they cannot always be effective and are certainly not a substitute for independent evidence, such as that obtained via crosses or transgenic experiments. Ultimately, association mapping is a powerful tool for identifying a list of candidates that is short enough to permit further genetic study.

Admixture as the basis for genetic mapping

“…A variety of different approaches have been developed to control statistically for genome-wide admixture and ancestry in modeling phenotypes [12,13,15,67]. At this relatively early stage in the development of methods, we need to know more about the utility of different approaches with different population and ancestry structures [12][13][14]16,66,67]. Empirical studies would be helpful, but there is a need for comprehensive modeling.…”

“…' Association mapping uses recombination events accumulated in natural populations over many generations to detect statistical associations between genetic loci of known genomic location and phenotypic traits of interest [11]. Association mapping generally requires the absence of population structure in the sample or statistical methods to account for structure [12][13][14][15]. Similarly, admixture Excess admixture quantifies the extent to which alleles and genotypes are introgressed into a foreign genetic background or are found in their native background.…”

“…The ability to account for population structure, in spite of the absence of controlled crosses or pedigrees, is a common feature of association and admixture mapping (Box 2). The absence of confounding population structure allows true, developmental genetic associations to be recognized, rather than the large number of spurious associations between allelic and phenotypic variation that are expected in the presence of population structure, where the effects of individual loci are not separated from their genetic background [12][13][14]16,17].…”

“…mouse, Arabidopsis, human), association mapping has become feasible at the scale of the entire genome, with a high likelihood of markers in linkage disequilibrium with genes underlying traits of interest (e.g. [14,81,82]). Consequently, both approaches now can involve mapping across the genome.…”

“…Finally, many of the models and methods developed for association mapping are relevant to admixture mapping. However, it is not clear that the population histories that have been modeled and investigated for association mapping encompass the considerable population structure and linkage disequilibrium that result from crosses between substantially diverged species and populations [12][13][14][15][16]66]. For example, more work remains to be done to compare the performance of different methods of analysis when population structure is great (e.g.…”