Assessing the prospects of genome-wide association studies performed in inbred mice
- 234 Downloads
The remarkable success in mapping genes linked to a number of disease traits using genome-wide association studies (GWAS) in human cohorts has renewed interest in applying this same technique in model organisms such as inbred laboratory mice. Unlike humans, however, the limited genetic diversity in the ancestry of laboratory mice combined with selection pressure over the past decades have yielded an intricate population genetic structure that can complicate the results obtained from association studies. This problem is further exacerbated by the small number of strains typically used in such studies where multiple spurious associations arise as a result of random chance. We sought to empirically assess the viability of GWAS in inbred mice using hundreds of expression traits for which the true location of the expression quantitative trait locus was known a priori. We then measured transcript abundance levels for these expression traits in 16 classical and 3 wild-derived inbred strains and carried out a genome-wide association scan, demonstrating the low statistical power of such studies and empirically estimating the large extent to which allelic association of transcripts gives rise to spurious associations. We provide evidence illustrating that in a large fraction of cases, the marker with the most significant p values fails to map to the location of the true eQTL. Finally, we provide experimental support for hundreds of traits, and that combining linkage analysis with association mapping provides significant increases in statistical power over a stand-alone GWAS as well as significantly higher mapping resolution than either study alone.
KeywordsQuantitative Trait Locus Association Mapping Inbred Strain Inbred Mouse Spurious Association
We thank The Jackson Laboratory, In Vivo Services, for their expert handling of the animals used in this study, and Rosetta Gene Expression Laboratory for the execution of the sample preparation and hybridization experiments. We also thank Eugene Chudin for insightful discourse on the methods and developing the computational tools that enabled this project. WS was supported in part by a fellowship from the Merck Research Laboratories. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The authors have declared that there are no conflicting interests.