R/readwrite.R

In kenhanscombe/ukbtools: Manipulate and Explore UK Biobank Data

#' Reads an Oxford format sample file
#'
#' This is a wrapper for \code{read_table} that reads an Oxford format .sample file. If you use the unedited sample file as supplied with your genetic data, you should only need to specify the first argument, file.
#'
#' @param file A path to a sample file.
#' @param col.names A character vector of column names. Default: c("id_1", "id_2", "missing")
#' @param row.skip Number of lines to skip before reading data.
#'
#' @seealso \code{\link{ukb_gen_read_fam}} to read a fam file
#'
#' @import readr
#' @export
#'
ukb_gen_read_sample <- function(
  file, col.names = c("id_1", "id_2", "missing"), row.skip = 2) {
  sample <- readr::read_table(file, skip = row.skip, col_names = col.names)
  as.data.frame(sample)
}




#' Reads a PLINK format fam file
#'
#' This is wrapper for read_table that reads a basic PLINK fam file. For plink hard-called data, it may be useful to use the fam file ids as a filter for your phenotype and covariate data.
#'
#' @param file A path to a fam file.
#' @param col.names A character vector of column names. Default: c("FID", "IID", "paternalID", "maternalID", "sex", "phenotype")
#' @param na.strings Character vector of strings to use for missing values. Default "-9". Set this option to character() to indicate no missing values.
#'
#' @seealso \code{\link{ukb_gen_read_sample}} to read a sample file
#'
#' @import readr
#' @export
#'
ukb_gen_read_fam <- function(
  file, col.names = c("FID", "IID", "paternalID", "maternalID", "sex", "phenotype"), na.strings = "-9") {
  fam <- readr::read_table(file, col_names = col.names, na = na.strings)
  as.data.frame(fam)
}




#' Writes a PLINK format phenotype or covariate file
#'
#' This function writes a space-delimited file with header, with the obligatory first two columns FID and IID. Use this function to write phenotype and covariate files for downstream genetic analysis in \href{https://www.cog-genomics.org/plink2}{plink} - the format is the same.
#'
#' @param x A UKB dataset.
#' @param path A path to a file.
#' @param ukb.variables A character vector of either the phenotypes for a PLINK phenotype file, or covariates for a PLINK covariate file.
#' @param ukb.id The id variable name (default = "eid").
#' @param na.strings String used for missing values. Defaults to NA.
#'
  #' @details The function writes the id variable in your dataset to the first two columns of the output file with the names FID and IID - you do not need to have two id columns in the data.frame passed to the argument \code{x}. Use the \code{--pheno-name} and \code{--covar-name} PLINK flags to select columns by name. See the PLINK documentation for the \code{--pheno}, \code{--mpheno}, \code{--pheno-name}, and \code{--covar}, \code{--covar-name}, \code{--covar-number} flags.
#'
#' @seealso  \code{\link{ukb_gen_read_sample}} to read a sample file, and \code{\link{ukb_gen_write_bgenie}} to write phenotype and covariate files to BGENIE format.
#'
#' @import dplyr readr
#' @importFrom magrittr "%>%"
#' @export
#' @examples
#' \dontrun{
#'
#' # Automatically inserts FID IID columns required by PLINK
#'
#' ukb_gen_write_plink(
#'    my_ukb_data,
#'    path = "my_ukb_plink.pheno",
#'    ukb.variables = c("height", "weight", "iq")
#' )
#'
#' ukb_gen_write_plink(
#'    my_ukb_data,
#'    path = "my_ukb_plink.cov",
#'    ukb.variables = c("age", "socioeconomic_status", "genetic_pcs")
#' )
#' }
#'
ukb_gen_write_plink <- function(x, path, ukb.variables, ukb.id = "eid", na.strings = "NA") {

  ids <- dplyr::transmute_(x, FID = ukb.id, IID = ukb.id)
  vars <- dplyr::select_(x, .dots = ukb.variables)

  readr::write_delim(cbind(ids, vars), path = path, na = na.strings, col_names = TRUE)
}




#' Writes a BGENIE format phenotype or covariate file.
#'
#' Writes a space-delimited file with a header, missing character set to "-999", and observations (i.e. UKB subject ids) in sample file order. Use this function to write phenotype and covariate files for downstream genetic analysis in \href{https://jmarchini.org/bgenie/}{BGENIE} - the format is the same.
#'
#' @param x A UKB dataset.
#' @param ukb.sample A UKB sample file.
#' @param ukb.variables A character vector of either the phenotypes for a BGENIE phenotype file, or covariates for a BGENIE covariate file.
#' @param path A path to a file.
#' @param ukb.id The eid variable name (default = "eid").
#' @param na.strings Character string to be used for missing value in output file. Default = "-999"
#'
#' @details Uses a \code{dplyr::left_join} to the sample file to match sample file order. Any IDs in the sample file not included in the phenotype or covariate data will be missing for all variables selected. See \href{https://jmarchini.org/bgenie-usage/}{BGENIE usage} for descriptions of the \code{--pheno} and \code{--covar} flags to read phenotype and covariate data into BGENIE.
#'
#' @seealso \code{\link{ukb_gen_read_sample}} to read a sample file, \code{\link{ukb_gen_excl_to_na}} to update a phenotype with NAs for samples to-be-excluded based on genetic metadata, and \code{\link{ukb_gen_write_plink}} to write phenotype and covariate files to PLINK format.
#'
#' @import dplyr readr
#' @importFrom magrittr "%>%"
#' @export
#' @examples
#' \dontrun{
#'
#' # Automatically sorts observations to match UKB sample file and writes missing values as -999
#'
#' my_ukb_sample <- ukb_gen_read_sample("ukb.sample")
#'
#' ukb_gen_write_bgenie(
#'    my_ukb_data,
#'    ukb.sample = my_ukb_sample,
#'    ukb.variables = c("height", "weight", "iq")
#'    path = "my_ukb_bgenie.pheno",
#' )
#'
#' ukb_gen_write_bgenie(
#'    my_ukb_data,
#'    ukb.sample = my_ukb_sample,
#'    ukb.variables = c("age", "socioeconomic_status", "genetic_pcs")
#'    path = "my_ukb_bgenie.cov",
#' )
#' }
#'
ukb_gen_write_bgenie <- function(
  x, ukb.sample, ukb.variables, path, ukb.id = "eid", na.strings = "-999") {

  names(ukb.sample)[1] <- ukb.id

  ukb.sample %>%
    dplyr::left_join(x, by = ukb.id) %>%
    dplyr::select_(.dots = ukb.variables) %>%
    readr::write_delim(path = path, na = na.strings, col_names = TRUE)
}