subset.metadata.R

#' Filter Metadata for Primary Cancer Site (Organ)
#' @description This function subsets the metadata for an organ. It works for meta data directly downloaded from TCGA or already filtered meta data. The subset is done by matching the "key" with the column names (patient ID or case_id) of the expression matrix.
#' @param meta_data Meta data provided by TCGA.
#' @param cancer_data A large list created prior by multimodalR for a cancer site.
#' @param key character - Unique identifier, default = "case_id"
#' @param additional_key needs to be added
#' @param additional_value needs to be added
#' @importFrom magrittr %>%
#' @examples
#' lungMetaData <- subset.metadata(metadata, lungXY)
#' @return Returns a data table of the metadata for a primary cancer site (organ).

subset.metadata <- function(meta_data, cancer_data, key = "case_id", additional_key = NULL, additional_value = NULL) {
  if(!is.character(meta_data[[key]])) {
  meta_data[[key]] <- as.character(meta_data[[key]]) #transform factor to character
  }
  correctRows <- numeric()                             #empty vector to be filled later
  caseNames <- names(cancer_data$Expressionmatrix)     #vector of case_id from expressionmatrix
  for (cases in caseNames) {                           #match case_id of the vector with case_id of metadata
  matchedRow <- grep(cases, meta_data[[key]])
  correctRows <- c(correctRows, matchedRow)
  }
  meta_data <- meta_data[c(correctRows),]              #subset by machted case_id
  if(!is.null(additional_key)) {
    additional_key <- rlang::sym(additional_key)

    meta_data <- meta_data %>% filter_(., !!(additional_key) == additional_value)
  }
  return(meta_data)
}