aml (version 0.1-1)

aml.pred.outside: Prediction With Adaptive Mixed LASSO

Description

This function is used to predict the genetic values for lines with marker information after fitting adaptive mixed LASSO on a training set using amltest.

Usage

aml.pred.outside(marker, response, kin, which.pred, numkeep, selectvar)

Arguments

marker

A matrix or data frame for the markers (or genetic effects). It should include both lines with observed phenotypic information (those in the training set) and those lines for which the genetic values will be predicted.

response

A numerical vector of trait (phenotype) values, corresponding to the lines in marker. For lines for which prediction will be made, the trait values can simply be set to NA. But for lines to be used in the training set, the trait value cannot be missing.

kin

The kinship matrix representing relationships between lines. It should correspond to the rows in marker and represent the relationships between lines in the training set as well as those to be predicted.

which.pred

A vector of integers specifying for which lines (which rows in marker) the prediction should be made. Lines not in which.pred will be used as the training set.

numkeep

This parameter is passed to amltest. It should be less than the number of lines in the training set.

selectvar

This parameter is passed to amltest. It should be less than numkeep.

Value

A list of the following:

predict.vl

The vector of predicted genetic values for lines specified in which.pred.

response.vl

The vector of observed phenotypic values for lines specified in which.pred. This is useful for cross-validation when comparing predicted and observed values. Otherwide it might be a vector of NAs.

Details

This function uses both marker effects and kinship to predict genetic values. Thus the kinship matrix should include both lines in the training set and the lines on which predictions are to be made. An adaptive mixed LASSO model is fitted for the training set including lines not in which.pred. The regression coefficients provided by amltest are then used for prediction. Besides performing prediction for lines with genetic marker genotypes but no phenotypic values, this function is especially convenient for performing cross-validation.

References

Wang, D., Eskridge, K.M. and Crossa, J. (2011) Identifying QTLs and Epistasis in Structured Plant Populations Using Adaptive Mixed LASSO. Journal of Agricultural, Biological, and Environmental Statistics, 16:170-184.

Wang, D., et al. (2012) Prediction of genetic values of quantitative traits with epistatic effects in plant breeding populations. Heredity, 109: 313-319.

See Also

amltest, aml.estimate.

Examples

     ## Predict the phenotype values of ten lines using the rest of the population in the wheat data
     data("wheat")
     clmarker<- cleanclust(wheat$marker, nafrac=0.2, mafb=0.1, corbnd=0.5, method="complete")
     intermat <- epigen(wheat$y, clmarker$newmarker, wheat$A, numkeep=100, selectvar=30, 
                        corbnd=0.5, mafb=0.04)
     which10<- sample(1:282, 10)
     pred10<- aml.pred.outside(intermat$effects, wheat$y, wheat$A, which10, 80, 40)