Function to solve univariate mixed models with or without the usage of omic information. This function allows single-step modeling of replicated observations with marker information available through the usage of a linkage function to connect to a whole-genome regression method. Genomic estimated values can be optionally deregressed (no shrinkage) while fitting the model.
mixed(y,random=NULL,fixed=NULL,data=NULL,X=list(),
alg=emML,maxit=10,Deregress=FALSE,...)The function wgr returns a list with Fitness values (Fitness) containing observation obs, fitted values hat, residuals res, and fitted values by model term fits; Estimated variance components (VarComp) containing the variance components per se (VarComponents) and variance explained by each model term (VarExplained), regression coefficients by model term (Coefficients), and the effects of structured terms (Structure) containing the marker effects of each model term where markers were provided.
Response variable from the data frame containg the dataset.
Formula. Right-hand side formula of random effects.
Formula. Right-hand side formula of fixed effects.
Data frame containing the response variable, random and fixed terms.
List of omic incidence matrix. Row names of these matrices connect the omic information to the levels of the indicated random terms (eg. X=list("ID"=gen)).
Function. Whole-genome regression algorithm utilized to solve link functions. These include MCMC (wgr, BayesB, etc) and EM (emEN, emDE, etc) algorithms. By default, it runs maximum likelihood emML.
Integer. Maximum number of iterations.
Logical. Deregress (unshrink) coefficients while fitting the model?
Additional arguments to be passed to the whole-genome regression algorithms especified on alg.
Alencar Xavier
The model for the whole-genome regression is as follows:
$$y = Xb + Zu + Wa + e$$
where \(y\) is the response variable, \(Xb\) corresponds to the fixed effect term, \(Zu\) corresponds to one or more random effect terms, \(W\) is the incidence matrix of terms with omic information and \(a\) is omic values by \(a=Mg\), where \(M\) is the genotypic matrix and \(g\) are marker effects. Here, \(e\) is the residual term. An example is provided using the data from the NAM package with: demo(mixedmodel).
Alterinative (and updated) implementations have similar syntax:
01) mm(y,random=NULL,fixed=NULL,data=NULL,
M=NULL,bin=FALSE,AM=NULL,it=10,verb=TRUE,
FLM=TRUE,wgtM=TRUE,cntM=TRUE,nPc=3)
02) mtmixed = function(resp, random=NULL, fixed=NULL,
data, X=list(), maxit=10, init=10, regVC=FALSE)
Xavier, A. (2019). Efficient Estimation of Marker Effects in Plant Breeding. G3: Genes, Genomes, Genetics, DOI: 10.1534/g3.119.400728
if (FALSE) {
demo(mixedmodel)
}
Run the code above in your browser using DataLab