rrBLUP-package: Ridge regression and other kernels for genomic selection
Description
This package has been designed for both genomic selection and association mapping. Some details of using the package for genomic selection have been published in the Plant Genome. The heart of the package is the function mixed.solve, which can be used to model marker effects as random effects or the genotypic values of the lines as random effects. In the latter case, the function A.mat is useful for calculating the additive relationship matrix and the prediction of breeding values. To include epistatic effects in the genotypic value predictions, use the function kinship.BLUP.Missing data
A number of improvements have been made concerning the handling of missing data since the original publication of the package. Both mixed.solve and kinship.BLUP will tolerate missing phenotypic data: those observations are simply not used. In addition, the package now has several features for handling missing genotypic data. For breeding value predictions, the function A.mat will calculate the additive relationship matrix using pairwise complete observations. This provides an unbiased estimate of the true A matrix but the calculation is more time consuming than if one imputes before calculating A. For small amounts of missing data, using impute to fill in with the population mean is a good option. For large amounts of missing data, I recommend using A.mat without imputing even though it takes longer. The function kinship.BLUP tolerates missing genotypic data for any of the three kernels.Association mapping
Use function GWA for association mapping.Parallel computing
For Mac, Linux, and UNIX users, R package multicore can be used in conjunction with rrBLUP to take advantage of multiple processors on a single machine. For large data sets, especially when there is missing data, I highly recommend trying out this feature, which is available with kinship.BLUP, A.mat, and GWA. You need R >= 2.14.1 for this to work properly, and you must also use R from the command line (not the GUI).References
Endelman, J.B. 2011. Ridge regression and other kernels for genomic selection with R package rrBLUP. Plant Genome 4:250-255.