rrBLUP-package: Ridge regression and other kernels for genomic selection

Description

This package has been developed primarily for genomic prediction with mixed models (but it can also do basic genome-wide association mapping with GWA). The heart of the package is the function mixed.solve, which is a general-purpose solver for mixed models with a single variance component other than the error. Genomic predictions can be made by estimating marker effects (RR-BLUP) or by estimating line effects (G-BLUP). In Endelman (2011) I made the poor choice of using the letter G to denotype the genotype or marker data. To be consistent with Endelman (2011) I have retained this notation in kinship.BLUP. However, that function has now been superseded by kin.blup and A.mat, the latter being a utility for estimating the additive relationship matrix (A) from markers. In these newer functions I adopt the usual convention that G is the genetic covariance (not the marker data), which is also consistent with the notation in Endelman and Jannink (2012).

Arguments

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. This is useful for Gaussian kernel predictions and when using the EM imputation algorithm in A.mat. 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. doi: 10.3835/plantgenome2011.08.0024

Endelman, J.B., and J.-L. Jannink. 2012. Shrinkage estimation of the realized relationship matrix. G3:Genes, Genomes, Genetics.