R/addDunn.R

In ACSoupir/Endophyte: Endophyte Summarizing Tool

#' Merges summarySYM with Dunn's Results
#'
#' @import stringr
#' @import plyr
#' @import FSA
#'
#' @author Alex Soupir
#'
#' @param data the raw data to be tested statistically
#' @param summaryData A data frame output by the summarySYM function
#' @param dunnComp the variable name in column 1 to
#' @param padj how to adjust the P value, if desired
#' @return The merged file of \code{summaryData} and FSA's dunnTest results
#' @export
#' @examples
#' mergeSYM(weightdata = plant_weight_data, winrhizo = root_winrhizo_data, cropname = "Plant Name)
#' merged_df <- mergeSYM(weightdata = plant_weight_data, winrhizo = root_winrhizo_data, cropname = "Plant Name)

addDunn <- function(data, summaryData, dunnComp, padj = NULL, tukey = NULL){
  x <- colnames(summaryData)[1]
  y <- colnames(summaryData)[3]

  i=1
  while(y != colnames(data)[i]){
    i=i+1
  }

  j=1
  while(x != colnames(data)[j]){
    j=j+1
  }

  if(is.null(padj)){
    padj = "none"
  }

  dunn <- dunnTest(data[[i]] ~ data[[j]], data, method = padj, two.sided = TRUE)[["res"]]

  df <- rename(data.frame(str_split_fixed(dunn$Comparison, " - ", 2)),
               c("X1"="X1","X2"="X2"))

  df2 <- data.frame(cbind(df,dunn))
  df3 <- df2

  df2.1 <- df2
  df2 = df2[df2$X1 == dunnComp, ]
  df2.1 = df2.1[df2.1$X2 == dunnComp, ]
  df2 = df2[,-1]
  colnames(df2)[1] = "X1"
  df2.1 = df2.1[,-2]
  df2 = rbind(df2,df2.1)
  df3 = df2[,c(1,4:5)]
  df3 = rename(df3,c(X1 = x))

  #  if (is.na(df2[1, 1]) == TRUE) {
  #    df3 = df3[df3$X2 == dunnComp, ]
  #    df3 = df3[, -c(2, 3, 4)]
  #    df3 <- rename(df3, c(X1 = x))
  #    t = 1
  #  }else {
  #    df3 = df3[df3$X1 == dunnComp, ]
  #    df3 = df3[, -c(1, 3, 4)]
  #    df3 <- rename(df3, c(X2 = x))
  #    t = 0
  #  }

  if(padj=="none"){
    df3 <- df3[,-c(3)]
    colnames(df3)[2] <- "Dunn.P.unadj"
  }else{
    colnames(df3)[2] <- "Dunn.P.unadj"
    df3 <- rename(df3, c("P.adj" = padj))
  }

  df4 <- merge(summaryData, df3, by.x=colnames(summaryData)[1], by.y=colnames(df3)[1], all.x=TRUE)

  if(!is.null(tukey)){
    ana <- aov(data[[i]] ~ data[[j]], data = data)
    tuk <- TukeyHSD(ana, ordered = TRUE)
    rst <- data.frame(tuk$`data[[j]]`)
    rst$Comparison <- rownames(rst)
    rst <- rst[,c(5,1:4)]

    rst.1 <- data.frame(str_split_fixed(rst$Comparison, "-", 2))

    rst.1$X1 <- as.character(rst.1$X1)
    rst.1$X2 <- as.character(rst.1$X2)

    for(k in 1:nrow(rst.1)){
      if(rst.1$X1[k] == dunnComp){
        rst.1$X3[k] = rst.1$X2[k]
      } else {
        rst.1$X3[k] = NA
      }
      if(rst.1$X2[k] == dunnComp){
        rst.1$X3[k] = rst.1$X1[k]
      }
    }

    rst.2 <- data.frame(cbind(rst.1$X3,rst$p.adj))

    colnames(rst.2)[2] <- "Tukey.p.adj"

    df4 <- merge(df4, rst.2, by.x=colnames(summaryData)[1], by.y=colnames(rst.2)[1], all.x=TRUE)
  }

  return(df4)
}

#height.ana <- aov(Height~TRT, data = gh)
#height.tukey <- TukeyHSD(height.ana, "TRT", ordered = TRUE)
#result <- data.frame(height.tukey$TRT)
#result["p.adj"]