predCrossVar

The predCrossVar package

contains a complete set of functions for the prediction of additive and dominance genetic variances and co-variances among full-siblings based on parents. predCrossVar enables the prediction of genetic variance on multi-trait selection indices. Built for diploid organisms with phased, chromosome- or linkage-group ordered biallelic marker data, and a centimorgan-scale genetic map.

Installation

You can install predCrossVar My GitHub with:

devtools::install_github("wolfemd/predCrossVar", ref = 'master')

HIGHLIGHTS

  • Allows for parents to be of arbitrary heterozygosity/homozygosity (outbred or inbred)
  • Predicts the additive and dominance genetic variances in the F1
  • Predicts genetic variances and co-variances. Enables prediction of genetic variance on an multi-trait selection index.
  • Handles simple (one trait, one cross) predictions, but built for complex (multi-trait, many crosses) scenarios.
  • Single estimate of marker effects from REML or MCMC (posterior mean effects –> predicts “variance of posterior means”) supported
  • Posterior Mean Variance (PMV) also supported: the estimator of Lehermeier et al. 2017b which computes the predicted variance across a sample of marker effects, e.g. the thinned MCMC samples, usually stored on disk. For the multi-trait case, a multivariate Bayesian model is required as only a marker effects for each trait must be computed on the same Gibbs chain.

Core functions

The package helps automate prediction for both simple (one trait, one cross) and complex (multi-trait, many crosses) scenarios.

Single-trait Functions REML or MCMCVPM
predCrossVarA() Predicts one cross variance
runCrossVarPredsA() Wraps around predCrossVarA() to predict multiple crosses

Equivalent functions for an additive-plus-dominance models are: predCrossVarAD() –> runCrossVarPredsAD() No function as of now to do dominance separately.

Note that these functions were developed early, and the multi-trait functions below should supersede these in function.

Multi-trait Functions REML, MCMCVPM, MCMCPMV
predOneCrossVarA() Predicts one cross, one variance (or covariance), additive-only model
predCrossVarsA() wraps around predOneCrossVarA(). Predicts multiple crosses (potentially in parallel/multicore), one variance (or covariance), additive-only model
runMtCrossVarPredsA() wraps around predCrossVarsA(). Predicts the variances and covariances in the multi-trait case for a set of crosses, additive-only model.

Equivalent functions for an additive-plus-dominance models are: predOneCrossVarAD() –> predCrossVarsAD() –> runMtCrossVarPredsAD(). No function as of now to do only dominance.

There are also functions for the much less computationally challenging prediction of cross means (predCrossMeanBVsOneTrait() –> predCrossMeanBVs()) for convenience.