Premium
Pedigree analysis package vs. MIXD: Fitting the mixed model on a large pedigree
Author(s) -
Olshen Adam B.,
Wijsman Ellen M.
Publication year - 1996
Publication title -
genetic epidemiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.301
H-Index - 98
eISSN - 1098-2272
pISSN - 0741-0395
DOI - 10.1002/(sici)1098-2272(1996)13:1<91::aid-gepi8>3.0.co;2-9
Subject(s) - markov chain monte carlo , statistics , gibbs sampling , mathematics , mixed model , heritability , monte carlo method , residual , variance (accounting) , markov chain , computer science , algorithm , bayesian probability , biology , accounting , genetics , business
Results of a simulation study with two methods of analysis of data simulated under themixed model on a 232‐member pedigree are presented. The programs Pedigree AnalysisPackage (PAP), which approximate the likelihoods needed in a complex segregation analysis,and MIXD, which uses Monte Carlo Markov chain (MCMC), to estimate likelihoods wereused. PAP obtained unbiased estimates of the major locus genotype means and the genefrequency, but biased estimates of the environmental variance component, and thus theheritability. A substantial fraction of the runs did not converge to an internal set of parameterestimates when analyzed with PAP. MIXD, which uses the Gibbs sampler to perform theMCMC sampling, produced unbiased estimates of all parameters with considerably moreaccuracy than obtained with PAP, and did not suffer from convergence of estimates to theboundary of the parameter space. The difference in behavior and accuracy of parameterestimates between PAP and MIXD was most apparent for models with either high or lowresidual additive genetic variance. Thus in situations where accuracy of the model isimportant, use of MCMC methods may be useful. In situations where less accuracy is needed,approximation methods may be adequate. Practical issues in using MCMC as implemented inMIXD to fit the mixed model are also discussed. Results of the simulations indicate that,unlike PAP, the starting configurations of most parameter estimates do not substantiallyinfluence the final parameter estimates in analysis with MIXD. © 1996 Wiley‐Liss,Inc.