Submitted to: The Plant Genome
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/28/2011
Publication Date: 6/21/2011
Citation: Hamblin, M., Jannink, J. 2011. Factors affecting the power of haplotype markers in association studies. The Plant Genome. 4:145-153.
Interpretive Summary: In genome-wide association studies DNA markers are correlated to a trait to identify markers that are close to genes that affect that trait (so called, quantitative trait loci, QTL). Simple markers that indicate the state of a single point in the genome can be used for this purpose but so can an adjacent series of simple markers. Such an adjacent series is called a haplotype. The question of whether simple markers or haplotypes are more powerful for identifying QTL is still unresolved. In this study, we simulated the process of genome evolution given different crop breeding histories and considering different QTL models. We then tested whether the QTL were identified more frequently with simple markers or haplotypes. We found that, across a range of plausible scenarios, grouping 2 or 3 simple markers gave greater QTL detection power. The average increase depended on the crop breeding history, the nature of the QTL itself, and on how the simple markers were developed, but the fact that haplotypes had greater power than simple markers was fairly robust. These results are particularly relevant to applications on elite breeding programs and when marker density is low.
Technical Abstract: An important, unresolved question in genome-wide association studies is whether there are predictable differences in power between single-SNP and haplotype markers. In this study, we use coalescent simulations to compare power for single-SNP and haplotype markers under a number of different models of demographic history and trait genetic architecture. We find that, across a range of plausible scenarios, the average power of 2- and 3-SNP haplotype markers to detect a QTL exceeds that of single SNP markers. The average increase in power is greater when a QTL is due to more than one polymorphism, when the population has experienced a bottleneck, and/or when marker SNPS are ascertained. These results are particularly relevant to applications where populations have experienced bottlenecks and marker density is low.