WGR2 (EM) {bWGR} | R Documentation |
Univariate models to find breeding values through regression fitted via expectation-maximization implemented in C++.
emRR(y, gen, df = 10, R2 = 0.5) emBA(y, gen, df = 10, R2 = 0.5) emBB(y, gen, df = 10, R2 = 0.5, Pi = 0.75) emBC(y, gen, df = 10, R2 = 0.5, Pi = 0.75) emBL(y, gen, R2 = 0.5, alpha = 0.02) emEN(y, gen, R2 = 0.5, alpha = 0.02) emDE(y, gen, R2 = 0.5) emML(y, gen, D = NULL) emCV(y, gen, k = 5, n = 5, Pi = 0.75, alpha = 0.02, df = 10, R2 = 0.5, avg=TRUE, llo=NULL, tbv=NULL)
y |
Numeric vector of response variable (n). |
gen |
Numeric matrix containing the genotypic data. A matrix with n rows of observations and m columns of molecular markers. |
df |
Hyperprior degrees of freedom of variance components. |
R2 |
Expected R2, used to calculate the prior shape (de los Campos et al. 2013). |
Pi |
Value between 0 and 1. Expected probability pi of having null effect (or 1-Pi if Pi>0.5). |
alpha |
Value between 0 and 1. Intensity of L1 variable selection. |
D |
NULL or numeric vector with length p. Vector of weights for markers. |
k |
Integer. Folding of a k-fold cross-validation. |
n |
Integer. Number of cross-validation to perform. |
avg |
Logical. Return average across CV, or correlations within CV. |
llo |
NULL or a vector (numeric or factor) with the same length as y. If provided, the cross-validations are performed as Leave a Level Out (LLO). This argument allows the user to predefine the splits. This argument overrides |
tbv |
NULL or numeric vector of 'true breeding values' (n) to use to compare cross-validations to. If NULL, the cross-validations will have the phenotypes as prediction target. |
The model for the whole-genome regression is as follows:
y = mu + Xb + e
where y is the response variable, mu is the intercept, X is the genotypic matrix, b is the effect of an allele substitution (or regression coefficient) and e is the residual term. A k-fold cross-validation for model evaluation is provided by emCV.
The EM functions returns a list with the intercept (mu), the regression coefficient (b), the fitted value (hat), and the estimated intraclass-correlation (h2).
The function emCV returns the predictive ability of each model, that is, the correlation between the predicted and observed values from k-fold cross-validations repeated n times.
Alencar Xavier
## Not run: data(tpod) emCV(y,gen,3,3) ## End(Not run)