copy of column harmonizaiton

R/CopyOfcolumn_harmization.R

@@ -0,0 +1,209 @@
+
+#' @title get_random_forest_model
+#' @param data_frame Mandatory, character \cr
+#'   Studyid number
+#' @param Liver_get_liver_om_lb_mi_tox_score_list Mandatory, character \cr
+#'   Studyid number
+#' @param not_Liver_get_liver_om_lb_mi_tox_score_list Mandatory, character \cr
+#'   path of database
+#' @return score
+#'
+#' @examples
+#' \dontrun{
+#' get_liver_lb_score(studyid='1234123', database_path = dbtoken)
+#' }
+#' @export
+#'
+
+
+
+get_harmonized_column <- function(data_frame = NULL, Liver_get_liver_om_lb_mi_tox_score_list,
+                                    not_Liver_get_liver_om_lb_mi_tox_score_list){
+# bismillah
+  `%ni%` <- Negate('%in%')
+  Impute <- T
+  ErrorMethod <- 'Prune' # Choose: "Flip" or "Prune" or "None"
+  Undersample <- T
+  reps <- 0
+  #rfDataRDS <- 'rfData_1000000_0.05_1_Prune_Round_8.rds'
+  threshold <- 0.05
+  holdback <- 1
+  testReps <- 5
+  testHoldBack <- 0.2
+  Round <- T
+  generateBarPlot <- T
+  nTopImportance <- 20
+  indeterminateUpper <- .75
+  indeterminateLower <- .25
+  Type = 1
+  hyperparameter_tuning <- F
+  removeEndpoints <- c('Infiltrate', 'UNREMARKABLE', 'THIKENING', 'POSITIVE')
+
+  ###-------------------------
+  # get lb score for Liver
+  Liver_master_LB_list <- Liver_get_liver_om_lb_mi_tox_score_list[['master_lb_score_six']]
+  Liver_master_LB_list$indst_TO <- "Liver"
+
+  # get LivertoBW score for Liver
+  Liver_master_liverToBW <- Liver_get_liver_om_lb_mi_tox_score_list[['master_liverToBW']]
+  Liver_master_liverToBW $indst_TO   <- "Liver"
+
+  # For mi score for Liver
+  Liver_master_MI_list <- Liver_get_liver_om_lb_mi_tox_score_list[['master_mi_df']]
+  Liver_master_MI_list$indst_TO <- "Liver"
+
+  # Reorder the columns to make `indst_TO` the first column
+  Liver_master_MI_list <- Liver_master_MI_list[, c("indst_TO", setdiff(names(Liver_master_MI_list), "indst_TO"))]
+  #~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+  # get lb score for not_Liver
+  not_Liver_master_LB_list <- not_Liver_get_liver_om_lb_mi_tox_score_list[['master_lb_score_six']]
+  not_Liver_master_LB_list$indst_TO  <- "not_Liver"
+
+  # get LivertoBW score for not_Liver
+  not_Liver_master_liverToBW <- not_Liver_get_liver_om_lb_mi_tox_score_list[['master_liverToBW']]
+  not_Liver_master_liverToBW$indst_TO  <- "not_Liver"
+
+  # For mi score for not_Liver
+  not_Liver_master_MI_list <- not_Liver_get_liver_om_lb_mi_tox_score_list[['master_mi_df']]
+  not_Liver_master_MI_list$indst_TO <- "not_Liver"
+
+  # Reorder the columns to make `indst_TO` the first column
+  not_Liver_master_MI_list <- not_Liver_master_MI_list[, c("indst_TO", setdiff(names(not_Liver_master_MI_list), "indst_TO"))]
+
+
+  #--------------------------
+  #~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+  LB <- rbind(Liver_master_LB_list, not_Liver_master_LB_list)
+
+  # Move the last column to the first position
+  LB <- LB[, c(ncol(LB), 1:(ncol(LB)-1))]
+
+
+  OM <- rbind(Liver_master_liverToBW, not_Liver_master_liverToBW)
+
+  # Move the last column to the first position
+  OM <- OM[, c(ncol(OM), 1:(ncol(OM)-1))]
+
+  ###------------------------
+  MIl <- Liver_master_MI_list
+  # Replace spaces and commas in column names with dots
+  colnames(MIl) <- gsub(' ', '.', colnames(MIl))
+  colnames(MIl) <- gsub(',', '.', colnames(MIl))
+  colnames(MIl) <- gsub('/', '.', colnames(MIl))
+
+
+  MIn <- not_Liver_master_MI_list
+  # Replace spaces and commas in column names with dots
+  colnames(MIn) <- gsub(' ', '.', colnames(MIn))
+  colnames(MIn) <- gsub(',', '.', colnames(MIn))
+  colnames(MIn) <- gsub('/', '.', colnames(MIn))
+
+
+
+  MIinter <- intersect(colnames(MIl), colnames(MIn))
+  MI <- rbind(MIl[, MIinter], MIn[, MIinter])
+
+  MIextraL <- setdiff(colnames(MIl), colnames(MIn))
+  MIextraN <- setdiff(colnames(MIn), colnames(MIl))
+  for (j in MIextraL) {
+    MI[, j] <- NA
+    MI[seq(nrow(MIl)), j] <- MIl[, j]
+  }
+  for (j in MIextraN) {
+    MI[, j] <- NA
+    MI[seq((nrow(MIl) + 1), nrow(MI)), j] <- MIn[, j]
+  }
+
+  MI[is.na(MI)] <- 0
+
+  #Identify Columns with Periods
+  findings2replaceIndex <- grep('.', colnames(MI), fixed = T)
+
+  # Store Column Names with Periods
+  f2replace <- colnames(MI)[findings2replaceIndex]
+
+  #Identify Unique Column Names without Periods
+  fn2replace <- unique(toupper(colnames(MI)[-findings2replaceIndex]))
+
+  #Remove Specific Columns from Processing
+  removeIndex <- which(fn2replace %in% c('INDST_TO',
+                                         'STUDYID',
+                                         'UNREMARKABLE',
+                                         'THIKENING',
+                                         'POSITIVE'))
+  fn2replace <- fn2replace[-removeIndex]
+
+  #Harmonize Synonym Columns
+  for (finding in fn2replace) {
+    synonyms <- grep(finding, f2replace, ignore.case = T, value = T)
+    for (synonym in synonyms) {
+      index <- which(MI[[synonym]] > 0)
+      for (i in index) {
+        if (MI[[synonym]][i] > MI[[finding]][i]) {
+          MI[[finding]][i] <- MI[[synonym]][i]
+        }
+      }
+    }
+  }
+
+  #Remove Synonym Columns
+  MI <- MI[,-findings2replaceIndex]
+
+  #return(list(LB,OM,MI))
+
+  ###------------------------
+  commonStudies <- Reduce(intersect, list(LB$STUDYID, OM$STUDYID, MI$STUDYID))
+  extraDomains <- c('OM', 'MI')
+  count <- 0
+  for (study in commonStudies) {
+    count <- count + 1
+    newRow <- LB[which(LB$STUDYID == study),]
+    for (domain in extraDomains) {
+      domainData <- get(domain)
+      studyIndex <- which(domainData$STUDYID == study)
+      for (j in colnames(domainData)[3:ncol(domainData)]) {
+        newRow[[j]] <- domainData[studyIndex, j]
+      }
+    }
+    if (count == 1) {
+      Data <- newRow
+    } else {
+      Data <- rbind(Data, newRow)
+    }
+  }
+
+
+  removeIndex <- which(colnames(Data) %in% removeEndpoints)
+  Data <- Data[, -removeIndex]
+
+  if (Round == T) {
+    zscoreIndex <- c(grep('avg_', colnames(Data)), grep('liver', colnames(Data)))
+    for (i in zscoreIndex) {
+      Data[, i] <- floor(Data[, i])
+      maxIndex <- which(Data[, i] > 5)
+      Data[maxIndex, i] <- 5
+    }
+    histoIndex <- which(substr(colnames(Data), 1, 1) %in% toupper(letters))
+    histoIndex <- histoIndex[-1]
+    for (i in histoIndex) {
+      Data[, i] <- ceiling(Data[, i])
+    }
+  }
+
+  columnSums <- sort(colSums(Data[,3:ncol(Data)], na.rm = T), decreasing = T)
+  Data[,3:ncol(Data)] <- Data[, names(columnSums)]
+  colnames(Data)[3:ncol(Data)] <- names(columnSums)
+
+
+
+   #write.csv(Data, 'mergedData.csv', row.names = F)
+
+
+
+
+print(rf)
+
+  return(rf)
+}
+
+