I would like to know of any publication studying the relative power of GWAS studies in different species. For example, I've seen reports that say genotyping and GWAS in dog breeds is much more powerful than in humans due to the extent of linkage disequilibrium and other population genetics considerations. Some claims say that many GWAS studies in dogs are 10-20x more powerful than many GWAS studies in humans. I have also heard from pigeons (pigeon-omics) to be a similar case. I would like to know if there are other comparisons for other species and the relative power of discovery for GWAS studies compared to human. For example, are there any numbers for yeast, Drosophila sp., C.elegans, zebrafish and then also for plants like maize, rice, etc.?
Part of the increase in power to detect a genetic association via GWAS-genotyping comes from long haplotypes. Many dog breeds went through selective breeding bottlenecks 100-200 years ago. Many lab model organisms, such as flies, worms and plants, have recombinant inbred lines that aid in GWAS discovery as well. The price one pays for this increased power, say for canine GWAS, is not being able to pinpoint the casual variant. The association SNP(s) found may be far, far away from the gene responsible for the phenotype, because of that long LD.
Another aspect not to neglect is the phenotype. In humans, there may be a 3-fold range in phenotype p, while in dog that value is just 35% (1.35x). That may mean that the phenotype is difficult to measure (accurately) and to assign to a given allele with statistical certainty.
A third aspect to consider is the effect size of the association, say on a per (risk) allele basis.
Thus, while I am not aware of any papers specifically taking such elements as haplotype length, phenotype range and effect size into an analysis of GWAS power across species, I feel that one versed in genetics of the organism of choice can readily accomplish this for the specific trait(s) of interest.
question looks like it's been dormant for a while, but i think there's some discussion to be had here-
I would argue that in many (most?) of the model organisms, power would be much greater than humans. Frequently (worms, mice, plants, yeast) you can work with basically isogenic inbred lines. I would argue this is much more important than long haplotypes: a) less importantly, no heterozygosity. b) more importantly, you can re-phenotype the same line repeatedly to directly estimate experimental/environmental variation, and get a very precise estimate of actual E(phenotype | genotype). this is opposed to humans, where you have the one individual, so you just have to pray that your experimental/environmental variation is low and your heritability is high in the sample population.
For example, for just 100 inbred lines of a plant, you get massive, beautiful GWAS peaks for many phenotypes: Atwell et al. 2010. Human GWAS sample sizes generally have to be in the 1000s before they are sufficiently powered (ref).
This subject is discussed a little further here.
In direct reference to the long haplotypes, note that power and mapping precision are different things. That is, your power to detect an association can be extremely high, but you might have a very broad chromosome interval that you will then have to go hunting in to find the causal locus. Note that this problem might actually be worse with large effects, which are probably under selection if they are interesting, and thus there is likely to be substantial linkage disequilibrium between the causal locus and surrounding regions. Of course, if the effect size is small (generally the case in humans though not in other orgs), this is less of an issue.