Association mapping in a simulated barley population

Abstract
The analysis of association mapping populations is frequently complicated by substructure within the population. If this is unaccounted for, it generally results in many false positive marker-traits associations. In this study, we simulate a large barley population, modelling inbreeding and selection, and compare five models for analysing marker-trait associations. One of these includes no population substructure, one includes information about geographical origin and type of barley to represent structure within the population and three use different approaches that have been proposed for estimating marker-based kinship. Kinship methods reduced the number of false positives substantially compared to the other models but none of these approaches had a clear advantage over the others. One solution is to fit more than one model and to consider as candidate associations those markers that are significant by all models. None of the approaches were very successful at detecting true associations at the lowest level of heritability considered, suggesting it is important to consider power in association mapping studies to avoid missing some true associations and overestimating others.
Year
2012
Category
Refereed journal
Output Tags
SG 2006-2011 WP 1.1 Barley Genetics