Make remove.bracket work better

i.e. more flexible to different potential
structures of the input data.frame.

Also allow the each.vs.ref columns to be
specified the other way around, which works better
if the reference sample is on the right e.g.
dose == 2.

R/add_pvalue.R

@@ -129,12 +129,16 @@ add_pvalue <- function(data,
 
   # check for remove.bracket
   # should stay before (!is.null(x))
-  if (remove.bracket) {
+  if (remove.bracket & !is.null(xmax)) {
     xmin.length <- length(unique(data[[xmin]]))
-    if (xmin.length == 1) {
+    xmax.length <- length(unique(data[[xmax]]))
+    if (xmin.length == 1 & xmax.length >= 2) {
       xmin <- xmax
       xmax <- NULL
     }
+    else if (xmin.length >= 2 & xmax.length == 1) {
+      xmax <- NULL
+    }
   }
 
   # determine the type of comparisons: one_group, two_groups, each_vs_ref, pairwise
@@ -155,8 +159,9 @@ add_pvalue <- function(data,
     }
     else if (ngroup1 >= 2 & ngroup2 >= 2) {
       comparison <- "pairwise"
-    } else {
-      stop("Make sure that xmin and xmax columns exist in the data.")
+    }
+    else if (ngroup1 >= 2 & ngroup2 == 1) {
+      comparison <- "each_vs_ref"
     }
   } else {
     if(ngroup1 >= 1) {
@@ -206,10 +211,6 @@ add_pvalue <- function(data,
 
   # draw brackets else draw p-value text
   if (pvalue.geom == "bracket") {
-    if (identical(data$xmin, data$xmax) | remove.bracket) {
-      bracket.size = 0
-    }
-
     params <- list(
       group = 1:nrow(data),
       label = "label", xmin = "xmin", xmax = "xmax",