Run k-fold cross-validation

Assess the accuracy of predicted previously unobserved genotypes (individuals) based on the available training data. Runs k-fold cross-validation for potentially multiple traits and optionally computing prediction accuracy on user-specified selection index. Three models are enabled: additive-only ("A"), additive-plus-dominance ("AD") and a directional-dominance model that incorporates a genome-wide homozygosity effect ("DirDom"). The union of all genotypes scored for all traits is broken into k-folds a user specified number of times. Subsequently each train-test pair is predicted for each trait and accuracies are computed.

runCrossVal(
  blups,
  modelType,
  selInd,
  SIwts = NULL,
  grms,
  dosages = NULL,
  nrepeats,
  nfolds,
  ncores = 1,
  nBLASthreads = NULL,
  gid = "GID",
  seed = NULL,
  ...
)

Arguments

blups: nested data.frame with list-column "TrainingData" containing BLUPs. Each element of "TrainingData" list, is data.frame with de-regressed BLUPs, BLUPs and weights (WT) for training and test.
modelType: string, "A", "AD", "DirDom". modelType="A": additive-only, GEBVS modelType="AD": the "classic" add-dom model, GEBVS+GEDDs = GETGVs modelType="DirDom": the "genotypic" add-dom model with prop. homozygous fit as a fixed-effect, to estimate a genome-wide inbreeding effect. obtains add-dom effects, computes allele sub effects (\(\alpha = a + d(q-p)\)) incorporates into GEBV and GETGV. "DirDom" requires dosages
selInd: logical, TRUE/FALSE, selection index accuracy estimates, requires input weights via SIwts
SIwts: required if selInd=FALSE, named vector of selection index weights, names match the "Trait" variable in blups
grms: list of GRMs where each element is named either A, D, or, AD. Matrices supplied must match required by A, AD and ADE models. For ADE grms=list(A=A,D=D)
dosages: dosage matrix. required only for modelType=="DirDom". Assumes SNPs coded 0, 1, 2. Nind rows x Nsnp cols, numeric matrix, with rownames and colnames to indicate SNP/ind ID
nrepeats: number of repeats
nfolds: number of folds
ncores: number of cores, parallelizes across repeat-folds
nBLASthreads: number of cores for each worker to use for multi-thread BLAS
gid: string variable name used for genotype ID's/ in e.g. blups (default="GID")
seed: numeric, use seed to achieve reproducibile train-test folds.
...

Value

Returns tidy results in a tibble with accuracy estimates for each rep-fold in a list-column "accuracyEstOut".

Arguments

Value

See also