gt_impute_xgboost: Imputation based XGBoost

Description

This function provides a simple imputation algorithm for gen_tibble objects based on local XGBoost models.

Usage

gt_impute_xgboost(
  x,
  alpha = 1e-04,
  size = 200,
  p_train = 0.8,
  n_cor = nrow(x),
  seed = NA,
  n_cores = 1,
  append_error = TRUE
)

Value

a gen_tibble with imputed genotypes

Arguments

x: a gen_tibble with missing data
alpha: Type-I error for testing correlations. Default is 1e-4.
size: Number of neighbour SNPs to be possibly included in the model imputing this particular SNP. Default is 200.
p_train: Proportion of non missing genotypes that are used for training the imputation model while the rest is used to assess the accuracy of this imputation model. Default is 0.8.
n_cor: Number of rows that are used to estimate correlations. Default uses them all.
seed: An integer, for reproducibility. Default doesn't use seeds.
n_cores: the number of cores to be used
append_error: boolean, should the xgboost error estimates be appended as an attribute to the genotype column of the gen_tibble. If TRUE (the default), a matrix of two rows (the number of missing values, and the error estimate) and as many columns as the number of loci will be appended to the gen_tibble. attr(missing_gt$genotypes, "imputed_errors")

Details

This function is a wrapper around bigsnpr::snp_fastImpute(). The error rates from the xgboost, if appended, can be retrieved with attr(x$genotypes, "imputed_errors") where x is the gen_tibble.

Examples

Run this code

if (FALSE) { # rlang::is_installed(c("xgboost", "RhpcBLASctl", "data.table"))
# \dontshow{
data.table::setDTthreads(2)
RhpcBLASctl::blas_set_num_threads(2)
RhpcBLASctl::omp_set_num_threads(2)
# }
example_gt <- load_example_gt("gen_tbl")

# Impute the gen_tibble
example_gt <- example_gt %>% gt_impute_xgboost()

# And we can check it has been imputed
example_gt %>% gt_has_imputed()
}