R/annotationfunctions.R

In FredHutch/tgR: TGR Convenience Functions

#' (Admin) Finds undefined REDCap variables
#'
#' Pulls sample data down from REDCap and compares them with the defined annotation list in GitHub and identifies variables in REDCap that need defining in GitHub.  Only variables for which there is a value in at least one record are returned.
#'
#' @return A data frame that is a template to add to the commonKnowledge data in the tgr-annotations repo with new definitions or corrections.
#' @author Amy Paguirigan
#' @details
#' Requires **admin** REDCap credentials to be set in the environment.
undefinedAnnotations <- function() {
  if ("" %in% Sys.getenv(c("TGR", "REDURI"))) {
    stop("You have missing environment variables.  Please set creds in env vars.")}
  else message("Credentials set successfully.")
  commonKnowledge <- tgrDefinitions()
  annotationsOnly <- commonKnowledge %>% dplyr::select(-c("Value", "ValueDescription", "bioScale", "centralDogma", "genomeScale", "DiseaseSpecific")) %>% unique()
  sciMeta <- tgrAnnotate(harmonizedOnly = F, DAG = "all", evenEmptyCols = T)
  # Remove project memberships
  defineMe <- sciMeta %>% dplyr::select(-dplyr::starts_with("data_is_"))
  categorical <- commonKnowledge %>% dplyr::filter(Type == "categorical")
  noLevelAnnots <- commonKnowledge %>% dplyr::filter(Type != "categorical")

  # Find all values used in the project, and find their associated Annotation
  usedAnnots <- purrr::map_dfr(colnames(defineMe), function(i){
    Y <- unique(dplyr::select(defineMe, tidyselect::all_of(i)))
    colnames(Y) <- "Value"
    Y$Value <- as.character(Y$Value)
    Y$Annotation <- i
    Y
  })
  #categorical annotations
  usedCat <- usedAnnots %>% dplyr::filter(Annotation %in% categorical$Annotation & is.na(Value) !=T )
  missingCat <- dplyr::anti_join(usedCat, categorical)
  missingCat <- dplyr::left_join(missingCat, annotationsOnly) %>% dplyr::mutate(ValueDescription = "")

  usedOther <- usedAnnots %>% dplyr::filter(!Annotation %in% categorical$Annotation) %>% dplyr::select(Annotation) %>% unique()
  missingOther <- dplyr::left_join(usedOther, commonKnowledge) %>% dplyr::filter(is.na(AnnotationDescription) == T) %>% dplyr::select(Annotation)
  missingOther <- dplyr::left_join(missingOther, annotationsOnly) %>% dplyr::mutate(Value = "", ValueDescription = "")

  makeMeaning <- rbind(missingCat, missingOther)

  #suppressWarnings(makeMeaning <- dplyr::bind_rows(commonKnowledge[0,], makeMeaning)) # make a sample data frame to fill in

  return(makeMeaning)
}

#' (Admin) Finds unused REDCap variables
#'
#' Pulls sample data down from REDCap and compares them with the defined annotation list in GitHub and identifies variables previously used in REDCap that have definitions in GitHub that need to be removed.
#'
#' @return A data frame that is list in the format of commonKnowledge data in the tgr-annotations repo with annotations and values to remove.
#' @author Amy Paguirigan
#' @details
#' Requires **admin** REDCap credentials to be set in the environment.
#' @export
unusedAnnotations <- function() {
  if ("" %in% Sys.getenv(c( "TGR", "REDURI"))) {
    stop("You have missing environment variables.  Please set creds in env vars.")}
  else message("Credentials set successfully.")
  commonKnowledge <- tgrDefinitions()
  sciMeta <- tgrAnnotate(harmonizedOnly = F, DAG = "all", evenEmptyCols = T)
  unused <- commonKnowledge %>% dplyr::filter(!Annotation %in% colnames(sciMeta)) %>% dplyr::filter(Category != "Bioinformatics")

  return(unused)
}