Premium
Accounting for uncertainty in QTL location in marker‐assisted pre‐selection of young bulls prior to progeny test
Author(s) -
STELLA A.,
JANSEN G. B,
BOETTCHER P. J.,
GIBSON J. P.,
LOHUIS M. M.,
PAGNACCO G.
Publication year - 2002
Publication title -
journal of animal breeding and genetics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.689
H-Index - 51
eISSN - 1439-0388
pISSN - 0931-2668
DOI - 10.1046/j.1439-0388.2002.00307.x
Subject(s) - bootstrapping (finance) , selection (genetic algorithm) , confidence interval , statistics , mathematics , herd , marker assisted selection , quantitative trait locus , biology , genetics , computer science , zoology , econometrics , machine learning , gene
The objective of this study was to evaluate whether the efficacy of marker assisted selection (MAS) could be improved by considering a confidence interval (CI) of QTL position. Specifically, MAS was applied for within‐family selection in a stochastic simulation of a closed nucleus herd. The location and effect of the QTL were estimated by least squares interval mapping with a granddaughter design and marker information was then used in a top down scheme. Three approaches were used to select the best bull within full sibships of 3 or 40 bulls. All three were based on the probability of inheriting the favorable allele from the grandsire (PROB). The first method selected the sib with the highest PROB at the location with the highest F‐ratio (MAX). The other two approaches were based on sums of estimated regression coefficients weighted by PROB at each cM within a 95% CI based on either bootstrapping (BOOT) or approximate LOD scores (LOD). Accounting for CI increased the relative genetic gain in all scenarios. The average breeding value (BV) of the selected bulls was increased by 2.00, 2.60 and 2.59% when MAS was applied using MAX, BOOT and LOD, respectively, compared to random selection (h 2 =0.30). Selected bulls carried the correct allele in 63.0, 68.5, 67.6 and 50.1% of the cases for MAX, BOOT, LOD and random selection, respectively.