Reliable computing in estimation of variance components

Journal of Animal Breeding and Genetics - Tập 125 Số 6 - Trang 363-370 - 2008
I. Misztal1
1University of Georgia, Athens, USA.

Tóm tắt

SummaryThe purpose of this study is to present guidelines in selection of statistical and computing algorithms for variance components estimation when computing involves software packages. For this purpose two major methods are to be considered: residual maximal likelihood (REML) and Bayesian via Gibbs sampling. Expectation‐Maximization (EM) REML is regarded as a very stable algorithm that is able to converge when covariance matrices are close to singular, however it is slow. However, convergence problems can occur with random regression models, especially if the starting values are much lower than those at convergence. Average Information (AI) REML is much faster for common problems but it relies on heuristics for convergence, and it may be very slow or even diverge for complex models. REML algorithms for general models become unstable with larger number of traits. REML by canonical transformation is stable in such cases but can support only a limited class of models. In general, REML algorithms are difficult to program. Bayesian methods via Gibbs sampling are much easier to program than REML, especially for complex models, and they can support much larger datasets; however, the termination criterion can be hard to determine, and the quality of estimates depends on a number of details. Computing speed varies with computing optimizations, with which some large data sets and complex models can be supported in a reasonable time; however, optimizations increase complexity of programming and restrict the types of models applicable. Several examples from past research are discussed to illustrate the fact that different problems required different methods.

Từ khóa


Tài liệu tham khảo

10.2527/2005.8361241x

Arango J., 2005, Study of codes of disposal at different parities of Large White sows using a linear censored model, J. Anim. Sci., 83, 2052, 10.2527/2005.8392052x

10.1016/j.livprodsci.2005.11.011

Blasco A., 2006, Philosophy of science and animal breeding research

BoxG.E.P DraperN.R.(1987) .Empirical Model‐Building and Response Surfaces Wiley.

10.1111/j.1439-0388.2008.00743.x

Druet T., 2006, Innovations in software packages in quantitative genetics

Foulley J.L., 2004, Prediction of random effects in linear mixed models under stochastic censoring, 129

10.1186/1297-9686-32-2-143

10.2527/1998.762441x

Gianola D., 2002, Likelihood, Bayesian and MCMC Methods in Quantitative Genetics

10.2307/2529975

10.3168/jds.S0022-0302(04)73436-0

10.1080/01621459.1996.10476714

10.1046/j.0931-2668.2003.00414.x

10.3168/jds.S0022-0302(90)78935-7

10.1016/S0301-6226(03)00003-4

10.1111/j.1439-0388.1989.tb00259.x

Meyer K., 2006, PX x AI: algorithmic for better convergence in restricted maximum likelihood estimation

10.1186/1297-9686-28-1-23

10.1111/j.1439-0388.1994.tb00473.x

Misztal I., 1999, Complex models, more data: simpler programming, Interbull Bull., 20, 33

10.3168/jds.S0022-0302(93)77478-0

Misztal I., 2007, A social competitive model with the categorical expression, J. Anim. Sci., 84, 416

Misztal I., 2002, BLUPF90 and related programs (BGF90)

10.1534/genetics.104.035956

10.2527/jas.2005-622

10.2527/2001.792333x

10.1186/1297-9686-39-2-123

10.3168/jds.S0022-0302(98)75707-8

Thompson R., 1999, Prospects for statistical methods in dairy cattle breeding, Interbull Bull., 20, 70

10.1098/rstb.2005.1676

10.1016/j.livprodsci.2003.09.016

10.3168/jds.S0022-0302(05)72782-X

10.2527/jas.2006-549

10.2527/jas.2006-550

10.2527/jas.2006-499

Thông tin
Thông tin xuất bản

Journal of Animal Breeding and Genetics

Tập 125 Số 6

363-370

Thông tin tác giả