Abstract
Quantitative trait loci (QTL) analysis is an effective tool for locating regions of the genome associated with a trait. Quantitative trait data are complex, and when statistically testing for the location of a QTL, the distribution of the test statistic is typically unknown. Historically, asymptotic thresholds have been difficult to derive for QTL analysis. Permutation testing has successfully provided significance thresholds for QTL analysis, but the need for exchangeability among the observations limits these empirically derived thresholds to simple linear models and does not permit the inclusion of important covariates in the model. We address the limitation of permutation theory for supplying empirically derived QTL significance threshold using a novel bootstrap threshold that is appropriate for multiple regression based interval mapping models. Simulation studies demonstrate that the proposed bootstrap thresholds improve detection and estimation of additive effects in QTL studies.
Keywords
Quantitative trait loci analysis, mixture models, permutation testing, bootstrap, multiple regression
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Recommended Citation
Ochsenfeld, Cherie A.; Jennings, Kristofer; and Doerge, R. W.
(2009).
"MAPPING QTL WITH COVARIATES,"
Conference on Applied Statistics in Agriculture.
https://doi.org/10.4148/2475-7772.1081
MAPPING QTL WITH COVARIATES
Quantitative trait loci (QTL) analysis is an effective tool for locating regions of the genome associated with a trait. Quantitative trait data are complex, and when statistically testing for the location of a QTL, the distribution of the test statistic is typically unknown. Historically, asymptotic thresholds have been difficult to derive for QTL analysis. Permutation testing has successfully provided significance thresholds for QTL analysis, but the need for exchangeability among the observations limits these empirically derived thresholds to simple linear models and does not permit the inclusion of important covariates in the model. We address the limitation of permutation theory for supplying empirically derived QTL significance threshold using a novel bootstrap threshold that is appropriate for multiple regression based interval mapping models. Simulation studies demonstrate that the proposed bootstrap thresholds improve detection and estimation of additive effects in QTL studies.
