R/DataPreparation.R

In A1exanderAlexeyuk/ariadne:

#' Unzip data
#'
#' @description unzipper function what unzip given data in directory given as an
#' argument
#'
#'
#'
#'
#' @param directoryZipFiles directory with zip files
#'
#' @param absolutePathDirectoryToCreate path to directory where to put unzipped files
#' if doesn't exist folder will be created
#' @export
#'
unzipper <- function(directoryZipFiles,
                     absolutePathDirectoryToCreate
){
  listOfFiles <- list.files(path = directoryZipFiles,
                            recursive = TRUE,
                            pattern = "\\.zip$",
                            full.names = T)
  lapply(listOfFiles, function(file) {
    folder <- gsub(".zip","", gsub(".*/","", file))

    dir.create(path = paste0(gsub("\\\\",
                                  '/',
                                  absolutePathDirectoryToCreate),
                             "/", folder), recursive = TRUE)

    utils::unzip(file, exdir = paste0(gsub("\\\\",
                                           '/',
                                           absolutePathDirectoryToCreate),
                                      "/", folder))
    return(NULL)
  })
}



#' Prepare data to KM plotting
#'
#' @param directories directories where stored time_to_event.csv
#'
#' @param targetIds target cohort ids to filter
#'
#' @param outcomeIds outcome cohort ids to filter
#'
#' @returns data.table dataframe
#'
#' @export
#'
#' @importFrom magrittr %>%
#' @examples
#' data <- kaplanMeierPlotPreparationDT(
#' targetIds = c(103,106),
#' outcomeIds = c(203, 212),
#' directories = dirs,
#' kaplanMeierDataCsv = "cohort_time_to_event.csv"
#' )
kaplanMeierPlotPreparationDT <- function(targetIds,
                                         outcomeIds,
                                         directories,
                                         kaplanMeierDataCsv = "cohort_time_to_event.csv") {
  unionAcrossDatabases <- lapply(directories, function(directory) {
    timeToEventTable <- data.table::fread(paste0(gsub("\\\\",
                                          '/',directory), "/", kaplanMeierDataCsv))
    return(timeToEventTable[target_id %in% targetIds & outcome_id %in% outcomeIds, ])


  })
  data.table::rbindlist(unionAcrossDatabases)
}


#' Prepare covariate data
#'
#' @param listOfDirectories directories where stored covariate.csv and covariate_value.csv
#'
#' @param filterWindowIds window_ids to filter
#'
#' @param cohortIds target cohort ids to filter
#'
#' @param covariateName name of csv contains covariate data (usually covariate.csv
#' or covariate_ref.csv)
#'
#' @param covariateValueName name of csv contains covariate data (usually covariate_value.csv
#'
#'
#' @returns data.table dataframe
#'
#' @importFrom magrittr %>%
#'
#' @importFrom data.table :=
#'
#' @export
#' @examples
#' data <- prepareCovariatesData(
#' listOfDirectories = directiries,
#' filterWindowIds = NULL,
#' cohortIds = c(103, 112),
#' covariateName = "covariate.csv",
#' covariateValueName = "covariate_value.csv"
#' )
prepareCovariatesData <- function(
                              listOfDirectories,
                              filterWindowIds = NULL,
                              cohortIds,
                              covariateName = "covariate.csv",
                              covariateValueName = "covariate_value.csv"
                              ) {
  listOfDF <- lapply(listOfDirectories, function(directory) {

    covariate <- data.table::fread(paste0(gsub("\\\\",
                                               '/',directory), "/", covariateName)) %>%
      data.table::setkey(covariate_id)
    covariateValue <- data.table::fread(paste0(gsub("\\\\",
                                                     '/',directory), "/", covariateValueName)) %>%
      data.table::setkey(covariate_id)


    covariatesForPlotting <- covariate[covariateValue, nomatch = NULL]

    covariatesForPlotting <- covariatesForPlotting[cohort_id  %in% cohortIds &
                                                     mean > 0,
                                                   ]
    if(!is.null(filterWindowIds)) {
      covariatesForPlotting[,
                            window_id := data.table::fcase(
                              covariate_id %% 10 == 4 , 4,
                              covariate_id %% 10 == 3 , 3,
                              covariate_id %% 10 == 2 , 2,
                              covariate_id %% 10 == 1 , 1
                            )]

      covariatesForPlotting <- covariatesForPlotting[window_id %in% filterWindowIds, ]
    } else {
      covariatesForPlotting
    }
  })
  return(data.table::rbindlist(listOfDF))
}

#' @export
#'
prepareFeatureProportionData <- function(listOfDirectories,
                                        filterWindowIds = NULL,
                                        cohortIds){
  listOfDF <- lapply(listOfDirectories, function(directory) {
    fp <- data.table::fread(paste0(gsub("\\\\",
                                               '/',directory), "/", "feature_proportions.csv"))
    featureProportionsConsolidated <- fp[
      cohort_id  %in% cohortIds & mean > 0,
    ]

    if(!is.null(filterWindowIds)) {
      featureProportionsConsolidated <- featureProportionsConsolidated[window_id %in%
                                                                         filterWindowIds,

      ]

    } else {
      featureProportionsConsolidated
    }
  })
  return(data.table::rbindlist(listOfDF))
}

#' Prepare covariate data for comparative analysis
#'
#' @param preparedCovariatesData csv or data table form prepareCovariatesData function
#'
#'
#' @param targetCogortId1 target cohort id 1
#'
#' @param targetCogortId2  target cohort id 2
#'
#' @param writeCsv boolean if TRUE write csv in working directory
#'
#'
#' @returns  merged data table with calculated SMD (SMD column)
#'
#' @importFrom data.table :=
#'
#' @export
#'
#' @examples
#'
#' dataF <- prepareCovariatesDataToPlotting(preparedCovariatesData = data,
#'cohortIds = c(103, 112)
#'
#'
prepareCovariatesDataToPlotting <- function(preparedCovariatesData,
                                            cohortIds,
                                            writeCsv = FALSE
) {

  dataToPlot <- preparedCovariatesData[cohort_id == cohortIds[2], ][
    preparedCovariatesData[cohort_id == cohortIds[1], ], on = .(
      covariate_id,
      database_id
      ), nomatch = NULL
  ] %>%
    data.table::setDT()

  dataToPlot[,
               SMD := (i.mean - mean)/
                 sqrt(i.mean * (1 - i.mean) +
                        (mean * (1 - mean)) / 2)
              ]
  if(writeCsv) {
    data.table::fwrite(
      x = dataToPlot,
      file = "PreparedCovariatesDataToPlotting.csv"
    )
  }
  return(dataToPlot)
}