ggplot2 aes_string() fails to handle names starting with numbers or containing spaces

Question

If the column names of a data.frame are started with numbers, or have spaces, aes_string() fails to handle them:

foo=data.frame(\"1         


        

Answer 1

I had similar situation and I used `` to make it understand:

mydf$ROIs <- row.names(mydf)
df.long <- melt(mydf)
colnames(df.long)[3] <- "Methods"
colnames(df.long)[4] <- "Mean L2 Norm Error"
variable.name="Variable", na.rm=TRUE)
ggplot(df.long, aes(ROIs, `Mean L2 Norm Error`, fill=Methods))+  
geom_bar(stat="identity",position="dodge")+
theme(axis.text.x=element_text(angle=45,hjust=1,vjust=1))

Answer 2

As far as I can tell, this method should work programmatically:

foo=data.frame("1st Col"=1:5, "2nd Col"=5:1, check.names=F)

#Save the colnames
bar=colnames(foo)

#change the names to something usable
names(foo) <- c("col1", "col2")

#Plot with arbitrary labs
ggplot(foo, aes(x=col1, y=col2)) + geom_point()+
  labs(x=bar[1], y=bar[2])

Answer 3

You can use the function aes_string2 below instead of aes_string:

aes_string2 <- function(...){
  args <- lapply(list(...), function(x) sprintf("`%s`", x))
  do.call(aes_string, args)
}

Answer 4

The original question asked how to modify a value of a variable so it is acceptable to ggplot() when the value is not know beforehand.

Write a function that adds back-ticks to the beginning and end of the value of the variable:

# ggName -> changes a string so it is enclosed in back-ticks.
#   This can be used to make column names that have spaces (blanks)
#   or non-letter characters acceptable to ggplot2.
#   This version of the function is vectorized with sapply.
ggname <- function(x) {
    if (class(x) != "character") {
        return(x)
    }
    y <- sapply(x, function(s) {
        if (!grepl("^`", s)) {
            s <- paste("`", s, sep="", collapse="")
        }
        if (!grepl("`$", s)) {
            s <- paste(s, "`", sep="", collapse="")
        }
    }
    )
    y 
}

Examples:

ggname(12)

[1] 12

ggname("awk ward")

"`awk ward`"

l <- c("awk ward", "no way!")
ggname(l)

"`awk ward`" "`no way!`"