Project Title:

Evaluating Methods of Updating Training Data in Long-Term Genomewide Selection


The objective of this project is to determine the dynamics of updating a training population and the long-term impact on a closed breeding program. We aimed to compare several methods of training population updating in terms of prediction accuracy, response to selection, and decay of genetic variance. This R package provide the code and data to run the simulation and create the figures used in the paper.


Please read our pre-print on biorXiv.

Our paper has been accepted at G3: Genes|Genomes|Genetics! You can read the early online paper, and the full version should be out in May.


To install, use the devtools package like so:


Obtaining Data

We used data from The Triticeae Toolbox data base in our simulations. A step-by- step set of instructions for obtaining and downloading this data is provided in a package vignette. Use the following code to access the vignette:


Running the Simulations

The simulations were run on a supercomputing platform that accepts batch submissions. The file inst/scripts/run_simulations_use.job is a Bash script that calls upon R and the script inst/scripts/run_experiment_use.R to run the simulations. Both of these scripts need to be edited (i.e. with different directory paths) before they can be used.

Replicating the Tables/Figures

Tables and figures can be replicated by using the procedure outlined in the Markdown file inst/scripts/plot_results_use.Rmd


Please open an issue for support with the package or to add comments.


Jeff Neyhart


To cite the pre-print, please use the following (formatting for Genetics):

Neyhart, J. L., T. Tiede, A. J. Lorenz, and K. P. Smith. 2016 Evaluating Methods of Updating Training Data in Long-Term Genomewide Selection. bioRxiv. doi:

UMN-BarleyOatSilphium/GSSimTPUpdate documentation built on May 9, 2019, 7:40 p.m.