benchmark_imputation: Benchmark mice imputation

In surh/HMVAR: Human Microbiome Variant Analysis in R

Description

Hide a percent of the observations and compare the results of imputation via mice with the original observations.

Usage

benchmark_imputation(
  geno,
  snp,
  outdir,
  p = 0.1,
  m = 5,
  verbose = FALSE,
  seed = NA,
  block_size = 100
)

Arguments

geno	A data table in BIMBAM format. Must contain one row per SNP, the first three columns must be site_id, minor_allele and major allele, followed by one column per sample. Missing values must be encoded as NA.
snp	A data table with snp information in BIMBAM format. Must contain columns ID, pos and chr in that order. Column ID must correspond to column site_id in geno.
outdir	Output directory to write the results.
p	Fraction of observations to hide.
m	Number of imputations performed. See mice documentation.
verbose	Whether to print progress on imputation.
seed	Seed for random sambling of data and for imputation if needed.
block_size	Number of markers to use to impute each marker.

Value

A list with elements r, p.imputed, res, and imputed_geno_file.

surh/HMVAR documentation built on Aug. 18, 2021, 1:21 a.m.