Plotting data by color variable with ggplot2 in R

Question

This is something of an extension of a previous question here:

Assign colors to a data frame based on shared values with a character string in R

I now have a data frame with x, y, errrs, and newcolors and I want to plot the data and error bars by color using ggplot2. I've tried this, and the plot works well, but the colors are not even close to being correct. I've tried defining the color variables in different places within the ggplot() call, but no luck. What am I missing?

Here is the data:

names          <- c( "TC3", "102", "172", "136", "142", "143", "AC2G" )
colors         <- c( "darkorange", "forestgreen", "darkolivegreen", "darkgreen", "darksalmon", "firebrick3", "firebrick1" )
dataA          <- c( "JR13-101A", "TC3B", "JR12-136C", "AC2GA", "TC3A" )
newcolors      <- rep( NA, length( dataA ) )
dataA          <- as.data.frame( cbind( dataA, newcolors ) )
x              <- c( 1, 2, 3, 4, 5 )
y              <- c( 10, 6, 3, 18, 2 )
errs           <- c( 2, 1, 2, 1, 2 )

dataA          <- cbind( dataA, x, y, errs )

and a solution to my previous question by @Dave2e that assigns colors by sample name:

dataA$newcolors <- as.character( dataA$newcolors )
 for( j in 1:length( names ) ) {
  dataA$newcolors[ grep( names[ j ], dataA$dataA ) ] <- colors[ j ] 
}

and finally the plotting code I've tried:

ggplot( dataA, aes( x = x, y = y) ) +
  geom_errorbar( aes( ymin = y - errs, ymax = y + errs, color = newcolors ), 
                 width = 0.03 ) + 
  geom_point( size = 5, aes( color = newcolors) )

(I've also tried putting the colors into the aes() call up front by aes( x = x, y = y, color = newcolors). The plot looks good, apart from the fact that the colors are not correct. "darkorange" shows up as a light green "darkgreen" is some pinkish color, and "firebrick1" is a light blue.

Maybe I'm missing something, but wouldn't it be easier to change the colors afterwards using scale_color_manual ? — lbusett
– lbusett, Commented Feb 25, 2017 at 16:32
Best to use scale_colour_manual or use colour = I(newColours) in the aes to use the colours as colours — Richard Telford
– Richard Telford, Commented Feb 25, 2017 at 16:40
Ahh, you are right @Richard Telford... Don't know how I missed the I(newcolors) tool. Thanks for clearing it up for me. — Jesse
– Jesse, Commented Feb 25, 2017 at 17:21
I don't quite understand how to use scale_color_manual to plot these, and the I(newcolors) call plots solid symbols, as opposed to symbols with a black border. How would I define these? — Jesse
– Jesse, Commented Feb 25, 2017 at 18:48

Community · Accepted Answer · 2017-05-23 12:16:51Z

In response to this comment of the OP, you may try

ggplot(dataA, aes(x = x, y = y, fill = I(newcolors))) +
  geom_errorbar(aes(ymin = y - errs, ymax = y + errs), 
                 width = 0.03) + 
  geom_point(size = 5, shape = 21)

In geom_point() size, shape and border color have been explicitely defined, i.e., outside of a call to aes(). Shapes numbered 21 to 24 are filled (see http://ggplot2.tidyverse.org/reference/scale_shape.html for available shapes). Consequently, the fill aesthetic but not the color aesthetic has been defined in the call to aes(). So, error bars and symbol borders are printed in black by default.

The advantage (or drawback, perhaps) is that the data point which had been assigned NA as newcolor is visible.

For comparison, below is how the plot looked after picking up the suggestion from Richard Telford's comment:

ggplot(dataA, aes(x = x, y = y, color = I(newcolors))) +
  geom_errorbar(aes(ymin = y - errs, ymax = y + errs), 
                width = 0.03) + 
  geom_point(size = 5)

Note that the leftmost data point was removed by ggplot2 as there was no color, i.e., NA, assigned to this data point.

Warning message:
Removed 1 rows containing missing values (geom_point).

Collectives™ on Stack Overflow

Plotting data by color variable with ggplot2 in R

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related