모체유전효과를 포함한 다형질 개체모형에서 신뢰도 추정 = Approximation of Accuracies for Breeding Values with Maternal Effects in Multiple Traits Animal Models

다국어 초록 (Multilingual Abstract)

The purpose of this study is to develop and evaluate an inexpensive method for approximating direct and maternal accuracy in very large models with combining multiple trait by ideas of Graser and Tier with single trait methodology of Misztal and Wiggans. The data set consisted of 16,380 Senepol growth records, with 22,136 animals in pedigree file. Traits included were birth weight, weaning weight and postweaning gain. The reliability of predictions is the squared correlation vetween true and predicted breeding values and the accuracy of evaluations is usually expressed in terms of squared root of reliability. We assume that accuracy by INV was desirable estimates even though inbreeding effects were ignored, the estimates by STA or MTA could be wrong or approximated. As comparing the accuracy estimates by STA to those by INV, those estimates were overestimated in direct genetic effects for birth weight with 0.08, direct genetic effects for weaning weight with 0.05, maternal genetic effects for birth weight with 0.07 and maternal genetic effect for weaning weight with 0.02. The accuracy estimates by STA were lower with 0.07 for postweaning gain than those by INV. MTA resulted in a further increase of the mean of all accuracy estimates with 0.06 for WWD to 0.14 for PGD. MTA is expected to increase the mean of all accuracy estimates because of extra information from correlated traits. Overestimation in MTA is most likely the result of overestimation in STA. Statistics on differences between INV and MTA approximations and correlations between accuracy estimates from INV and STA and MTA. The MTA were from 0.02 to 0.11 different, on average, and the Standard Deviation(SD) is up to 0.05. The maternal effects had lower SD. Accuracy estimates by MTA were better for direct traits than those by STA, by up to 0.04. The small decrease was found for maternal effects. Most of the accuracy estimates are in the lower range because of relatively small data set used for this study. The results of comparison are strongly affected by variance components used in the model. Superiority of MTA over STA would increase with larger correlations among traits. Accuracy of STA would increase of the correlation between the maternal and direct effects were small.