Selecting Informative Traits for Multivariate Quantitative

20 A major consideration in multitrait analysis is which traits should be jointly analyzed. As a 21 common strategy, multitrait analysis is performed on either pairs of traits or on all of traits. To 22 fully exploit the power of multitrait analysis, we propose variable selection to choose a subset of 23 informative traits for multitrait quantitative trait locus (QTL) mapping. The proposed method 24 is very useful for achieving optimal statistical power for QTL identification and for disclosing 25 the most relevant traits. It is also a practical strategy to effectively take advantage of multitrait 26 analysis when the number of traits under consideration is too large, making the usual multi27 variate analysis of all traits challenging. We study the impact of selection bias and the usage of 28 permutation tests in the context of variable selection, and develop a powerful implementation 29 procedure of variable selection for genome scanning. We demonstrate the proposed method and 30 selection procedure in a backcross population using both simulated and real data. The extension 31 to other experimental mapping populations is straightforward. 32