fix plotLoadings error caused by long variable names. Closes #45

R/plotLoadings.mint.spls.R

@@ -173,14 +173,16 @@ xlim = NULL,
 
             #display barplot with names of variables
             #added condition if all we need is the contribution stats
+            colnames.X <- .trim_long_names(colnames.X) ## issue 45
+
             if (!is.null(title) & length(block) > 1)
             {
                 par(mar = c(4, max(7, max(sapply(colnames.X, nchar),na.rm = TRUE)/2), 6, 2))
             } else {
                 par(mar = c(4, max(7, max(sapply(colnames.X, nchar),na.rm = TRUE)/2), 4, 2))
             }
 
-            mp = barplot(df$importance, horiz = TRUE, las = 1, col = df$color, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
+            barplot(df$importance, horiz = TRUE, las = 1, col = df$color, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
             cex.names = size.name, cex.axis = 0.7, beside = TRUE, border = border, xlim = xlim)
 
             if ( length(block) == 1 & is.null(title) )

R/plotLoadings.mint.splsda.R

@@ -230,14 +230,16 @@ xlim = NULL,
 
             #display barplot with names of variables
             #added condition if all we need is the contribution stats
+            colnames.X <- .trim_long_names(colnames.X) ## issue 45
+
             if (!is.null(title) & length(block) > 1)
             {
                 par(mar = c(4, max(7, max(sapply(colnames.X, nchar),na.rm = TRUE)/2), 6, 2))
             } else {
                 par(mar = c(4, max(7, max(sapply(colnames.X, nchar),na.rm = TRUE)/2), 4, 2))
             }
 
-            mp = barplot(df$importance, horiz = TRUE, las = 1, col = df$color, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
+            barplot(df$importance, horiz = TRUE, las = 1, col = df$color, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
             cex.names = size.name, cex.axis = 0.7, beside = TRUE, border = border, xlim = xlim[i, ])
 
             if ( length(block) == 1 & is.null(title) )

R/plotLoadings.pca.R

@@ -72,9 +72,10 @@ xlim = NULL,
     df = data.frame(importance = value.selected.var) # contribution of the loading
 
     # barplot with contributions
+    colnames.X <- .trim_long_names(colnames.X) ## issue 45
     par(mar = c(4, max(7, max(sapply(colnames.X, nchar),na.rm = TRUE)/3), 4, 2))
 
-    mp = barplot(df$importance, horiz = TRUE, las = 1, col = col, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
+    barplot(df$importance, horiz = TRUE, las = 1, col = col, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
     cex.names = size.name, cex.axis = 0.7, beside = TRUE, border = border, xlim = xlim)
 
     if (is.null(title))

R/plotLoadings.spls.R

@@ -83,14 +83,16 @@ xlim = NULL,
         df = data.frame(importance = value.selected.var) # contribution of the loading
 
         # barplot with contributions
+        colnames.X <- .trim_long_names(colnames.X) ## issue 45
+
         if (!is.null(title) & length(block) > 1)
         {
             par(mar = c(4, max(7, max(sapply(colnames.X, nchar),na.rm = TRUE)/3), 6, 2))
         } else {
             par(mar = c(4, max(7, max(sapply(colnames.X, nchar),na.rm = TRUE)/3), 4, 2))
         }
 
-        mp = barplot(df$importance, horiz = TRUE, las = 1, col = col, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
+        barplot(df$importance, horiz = TRUE, las = 1, col = col, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
         cex.names = size.name, cex.axis = 0.7, beside = TRUE, border = border, xlim = xlim[i,])
 
         if ( (length(block) == 1 & is.null(title)) | (length(block) > 1 & missing(subtitle)))

R/plotLoadings.splsda.R

@@ -154,6 +154,7 @@ xlim = NULL,
             }
 
             # display barplot with names of variables
+            colnames.X <- .trim_long_names(colnames.X) ## issue 45
             if (plot) # condition if all we need is the contribution stats
             {
                 if (!is.null(title) & length(block) > 1)
@@ -163,7 +164,7 @@ xlim = NULL,
                     par(mar = c(4, max(7, max(sapply(colnames.X, nchar), na.rm = TRUE)/3), 4, 2))
                 }
 
-                mp = barplot(df$importance, horiz = TRUE, las = 1, col = df$color, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
+                barplot(df$importance, horiz = TRUE, las = 1, col = df$color, axisnames = TRUE, names.arg = colnames.X, #names.arg = row.names(df),
                 cex.names = size.name, cex.axis = 0.7, beside = TRUE, border = border, xlim = xlim[i, ])
 
                 if ( length(block) == 1 & is.null(title) )

R/utils.R

@@ -0,0 +1,29 @@
+## ----------- .trim_long_names ----------- 
+#' Trim long variable names
+#'
+#' Trims feature names longer than \code{len} adding \code{...} at the end of
+#' the name with a message. This helps avoid margin errors in plots. 
+#' \code{len=23} is chosen as default only because longest ensemble id for
+#' mouse/human genes/transcripts is this length.
+#' 
+#' @param var_names character vector of variable names
+#' @param len integer, names longer than this will be trimmed so total length
+#' will be \code{len}, including \code{...}.
+#'
+#' @return character vector of trimmed names
+#'
+#' @examples
+#' .trim_long_names(var_names = c("long-variable-name-of-length-31", "short-variable"))
+#' @noRd
+.trim_long_names <- function(var_names=c("long-variable-name-of-length-31", "short-variable"), len=23) {
+    if (any(nchar(var_names) > len)) {
+        message(sprintf("Some variable names are too long. Trimmed for visualisation purposes."))
+
+        var_names <- sapply(var_names, function(char) {
+            ifelse(nchar(char) > len, sprintf("%s...", substring(char, 1, len-3)) , char)
+        })
+    }
+
+    return(var_names)
+}
+