13  Genomic Prediction

13.1 Internal GBLUP (Known h²)

pop <- breeding.diploid(
  pop,
  bve = TRUE,
  bve.gen = 1:5,
  heritability = 0.5     # Must provide h²
)

13.2 rrBLUP Package

pop <- breeding.diploid(
  pop,
  bve = TRUE,
  bve.gen = 1:5,
  bve.rrblup = TRUE     # Uses rrBLUP::mixed.solve
)

13.3 Bayesian Methods (BGLR)

pop <- breeding.diploid(
  pop,
  bve = TRUE,
  bve.gen = 1:5,
  bve.bglr = TRUE,
  bglr.model = "BayesB"   # BayesA, BayesB, BayesC, etc.
)

13.4 BLUPF90

pop <- breeding.diploid(
  pop,
  bve = TRUE,
  bve.gen = 1:5,
  bve.blupf90 = TRUE,
  blupf90.path = "/path/to/blupf90"
)

13.5 Cross-Validation

# Assess prediction accuracy
accuracy <- analyze.bv(
  pop,
  gen = 5,
  cohorts = "TestSet",
  bve.gen = 1:4         # Training set
)

print(accuracy$correlation)  # Accuracy
print(accuracy$bias)          # Bias

13.6 Summary

  • Multiple BVE methods
  • Integration with R packages and external software
  • Accuracy assessment

Continue to Chapter 14: Population Structure!