Question

This question is related to the one I asked here. I’m trying to color individual points in a scatterplot with a color specified in a column of the dataframe. The data is from a tab delimited file and looks like this:

> dput(ztestHSV)
structure(list(y = c(0, -1, -1, -2), x = c(0, 2, 0, -2), group = c("m", 
"m", "m", "s"), colorHSV = c("0.02,0.83,0.89", "0.59,0.59,0.85", 
"0.25,0.45,0.8", "0.55,0.41,0.8"), colorText = c("red", "blue", 
"green", "turquoise")), .Names = c("y", "x", "group", "colorHSV", 
"colorText"), class = "data.frame", row.names = c(NA, -4L))

i.e.

   y  x group  colorHSV       colorText
1  0  0     m  0.02,0.83,0.89       red
2 -1  2     m  0.59,0.59,0.85      blue
3 -1  0     m  0.25,0.45,0.8      green
4 -2 -2     s  0.55,0.41,0.8  turquoise

The following code works well for the predefined set of colors available in R:

ztestHSV=read.table(file="testHSV.txt",sep="\t",header=TRUE,
                    colClasses=c("numeric","numeric","character","character","character"))

plot(ztestHSV[ztestHSV$group=='m',]$x,ztestHSV[ztestHSV$group=='m',]$y,type='n')
points(ztestHSV[ztestHSV$group=='m',]$x,ztestHSV[ztestHSV$group=='m',]$y,
       pch=23,cex=1.5,
       bg=ztestHSV[ztestHSV$group=='m',]$colorText
)

What I would really like to do is use hsv() to specify the colors. But I can’t figure out the right bg command to make it work. I’ve tried several things and listed them below with the associated result.

1

bg=hsv(as.numeric(ztestHSV$colorHSV))

Error in hsv(as.numeric(ztestHSV$colorHSV)) : 
bad hsv to rgb color conversion
In addition: Warning message:
In hsv(as.numeric(ztestHSV$colorHSV)) : NAs introduced by coercion

2

bg=hsv(as.numeric(strsplit(ztestHSV$colorHSV,",")))

Error in hsv(as.numeric(strsplit(ztestHSV$colorHSV, ","))) : 
(list) object cannot be coerced to type 'double'

3

bg=hsv(unlist(strsplit(ztestHSV$colorHSV,",")))

Above runs but uses the wrong background colors

4

bg=hsv(as.numeric(unlist(strsplit(ztestHSV$colorHSV,","))))

Above doesn’t generate an error but doesn’t fill background colors

It seems like this is a problem with converting the character string of colorHSV to something that hsv() can use, but I don’t know how to do it. Any help would be great.

Answer 1

It helps if you provide an easily reproducible example by using something like dput() to show the data you are using:

> dput(ztestHSV)
structure(list(y = c(0L, -1L, -1L, -2L), x = c(0L, 2L, 0L, -2L
), group = structure(c(1L, 1L, 1L, 2L), .Label = c("m", "s"), class = "factor"), 
    colorHSV = c("0.02,0.83,0.89", "0.59,0.59,0.85", "0.25,0.45,0.8", 
    "0.55,0.41,0.8"), colorText = structure(c(3L, 1L, 2L, 4L), .Label = c("blue", 
    "green", "red", "turquoise"), class = "factor")), .Names = c("y", 
"x", "group", "colorHSV", "colorText"), row.names = c(NA, -4L
), class = "data.frame")

You can convert the vector of character strings into a list of numerics, and then use do.call() to call hsv() on each one, like this:

> bg <- sapply(strsplit(ztestHSV$colorHSV, ","), function(x) do.call(hsv, as.list(as.numeric(x))))
> dput(bg)
c("#E33D27", "#5994D9", "#9ECC70", "#78B3CC")
> plot(1:4, pch = 21, cex = 20, bg = bg)