Practical implementation of a breeding program with genomic selection

In this part of the vignette, we will implement the genomic selection in a Maize pipeline. Two scenarios will be implemented after a fifteen-year Burn-In period. The first scenario will account only for truncated phenotypic selection. In the second scenario, we will implement genomic selection in the pipeline. So, in year 4, the individuals will be selected based on the estimated breeding values that came from a genomic model (ridge regression model). In addition, the parents will be selected based on ‘ebv’.

Package and files for recording the outputs

rm(list=ls())

# Packages
require(AlphaSimR)

# Creating the files to record the results
MeanG_pop = matrix(NA, 35)
MeanA_pop = matrix(NA, 35)
VarG_pop  = matrix(NA, 35)

Create parents and fill the pipeline

# Parameters files
source("aux_files/GlobalParameters.R")

# Create the parents
source("aux_files/Create_parents.R")

# FillPipeline
source("aux_files/FillPipeline.R")

# Environmental covariate
P = runif(burninYears+futureYears)

Burn-In period

for(year in 1:burninYears){
p = P[year]
cat("Working on year:",year,"\n")

source("aux_files/UpdateParents.R")  # Pick new parents based on last year's data
source("aux_files/UpdateTesters.R")  # Pick new testers and hybrid parents
source("aux_files/AdvanceYearPS.R")  # Advances yield trials by a year
source("aux_files/WriteRecordsGS.R") # Write records for GS predictions
source("aux_files/UpdateResults.R")  # Track summary data

}

# Saving the info

save.image("results/BURNIN.RData")

Phenotypic selection program

# 3.0 Loading the scenarios
load("results/BURNIN.RData")


for(year in (burninYears+1):(burninYears+futureYears)){
p = P[year]

# 3.1 Loop
cat("Working on year:",year,"\n")
source("aux_files/UpdateParents.R")  # Pick new parents based on last year's data
source("aux_files/UpdateTesters.R")  # Pick new testers and hybrid parents
source("aux_files/AdvanceYearPS.R")  # Advances yield trials by a year
source("aux_files/UpdateResults.R")  # Track summary data
  
}

# 3.2 Recording results
output1 = data.frame(scenario=rep("PS", 35), # Scenario name
                     MeanG_pop,
                     MeanA_pop,
                     VarG_pop,
                     stringsAsFactors=FALSE)

# 3.3 Saving the results as RDS
saveRDS(output1,"results/Results_PS.rds")

Genomic selection program

# 3.0 Loading the scenarios
load("results/BURNIN.RData")


for(year in (burninYears+1):(burninYears+futureYears)){
p = P[year]

cat("Working on year:",year,"\n")
if(year == (burninYears+1)){
  source('aux_files/fillGS.R')
}
source("aux_files/UpdateParents_GS.R")  # Pick new parents based on last year's data
source("aux_files/UpdateTesters.R")  # Pick new testers and hybrid parents
source("aux_files/AdvanceYearGS.R")  # advance the populations
source("aux_files/WriteRecordsGS.R")  # Track records for GS
source("aux_files/UpdateResults.R")  # Track summary data
  
}

# 3.2 Recording results
output1 = data.frame(scenario=rep("GS", 35), # Scenario name
                     MeanG_pop,
                     MeanA_pop,
                     VarG_pop,
                     stringsAsFactors=FALSE)

# 3.3 Saving the results as RDS
saveRDS(output1,"results/Results_GS.rds")

Plotting the results

# Loading the results
Scenario1 = readRDS("results/Results_PS.rds")
Scenario2 = readRDS("results/Results_GS.rds")


# Plot hybrid mean of genetic values over time
meanRanges = range(c(Scenario1$MeanG_pop[15:35], Scenario2$MeanG_pop[15:35]))
plot(x = 15:35, y = Scenario1$MeanG_pop[15:35], type = "l", col = "black", lwd = 3,
     xlab = "Year", ylab = "Mean of genetic values", ylim = meanRanges)
lines(x = 15:35, y = Scenario2$MeanG_pop[15:35], type = "l", col = "blue", lwd = 3)
legend(x = "topleft", legend = c('Pheno', 'GS'), title = "Scenarios",
       lwd = 3, lty = c(1, 1), col = c("black", "blue"), bty = "n")

References

  1. University of Florida, ↩︎

  2. University of Florida, ↩︎