Circular plot in ggplot2 with line segments connected in r

Question

I am trying to create a circular plot and am stuck at a point:

dat1 <- data.frame (xvar = 1:10, y = 6, ymin = 4, ymax = 4.5)

Using this data, I can produce a circular ribbon plot in ggplot2

require(ggplot2) 
 ggplot(dat1, aes(x=xvar, y=y)) +  geom_ribbon(aes(ymin=ymin, ymax=ymax),
  col = "blue", fill = "blue2") + ylim (c(0,6)) + coord_polar()

However I want more.

I want to fill the segment of the ribbon with different colors and labels using the following data.

 filld <- data.frame (start = c(1, 4, 6, 7.5, 8, 9), end = c(4, 6, 7.5, 8, 9, 10),
                         label = c("A", "B", "C", "A", "C", "D"))
  filld
##    start  end label
## 1   1.0  4.0     A
## 2   4.0  6.0     B
## 3   6.0  7.5     C
## 4   7.5  8.0     A
## 5   8.0  9.0     C
## 6   9.0 10.0     D

The ribbon will be filled with different color by label variable. For example, the segment A will start from 1 and end at 4. Then segment B will start and end at 6 and filled with different color. Segments with same label (such as A and C) will be connected by line.

The resulting plot will look like this:

Nice graphs :) It's great to add anti-aliasing

Viet
– Viet

2012-07-25 02:29:52 +00:00
Commented Jul 25, 2012 at 2:29 — Viet
– Viet, Commented Jul 25, 2012 at 2:29

BSMP · Accepted Answer · 2018-12-21 13:37:56Z

19

Here is an example:

filld$p <- rowMeans(subset(filld, select = c(start, end)))
ggplot(filld, aes(xmin = start, xmax = end, ymin = 4, ymax = 5, fill = label)) + 
  geom_rect() + 
  geom_segment(data = subset(filld, label %in% label[duplicated(label)]),
               aes(x = p, y = 0, xend = p, yend = 4, colour = label),
               size = 2, show_guide = FALSE) +
  geom_text(aes(x = p, y = 4.5, label = label), colour = "white", size = 10) +
  coord_polar() + 
  scale_y_continuous(limits = c(0, 5))

Updated

I do not recommend but something like this:

filld <- data.frame (start = c(1, 4, 6, 7.5, 8, 9), end = c(4, 6, 7.5, 8, 9, 10),
                     label = c("A", "B", "C", "A", "C", "D"))
filld$p <- rowMeans(subset(filld, select = c(start, end)))
filld <- merge(filld, ddply(filld, .(label), summarize, p2 = mean(p)))

lnd <- subset(filld, label %in% label[duplicated(label)])
lnd <- ddply(lnd, .(label), function(x) {
  x <- seq(x$p[1], x$p[2], length = 100)
  y <- 4.5 + ((x - mean(x))^2 - (x[1]-mean(x))^2) / (x[1]-mean(x))^2 * 3 + sin(x*3*pi) * 0.1
  data.frame(x, y)
})


p <- ggplot(filld, aes(xmin = start, xmax = end, ymin = 4, ymax = 5, colour = label, fill = label)) + 
  geom_line(aes(x, y, xmin = NULL, ymin = NULL, xmax = NULL, ymax = NULL), data = lnd, size = 2) +
  geom_rect() + 
  geom_text(aes(x = p, y = 4.5, label = label), colour = "white", size = 10) +
  coord_polar() + 
  scale_y_continuous(limits = c(0, 5))
p

Perhaps, what you want is beyond the scope of ggplot2.

edited Dec 21, 2018 at 13:37

BSMP

4,9008 gold badges36 silver badges46 bronze badges

answered Jul 25, 2012 at 2:39

kohske

67.2k9 gold badges168 silver badges155 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

shNIL Over a year ago

@kohske pretty nice ...lines are much better, but I wonder if it is possible make lines that are lower than 180 to making smaller loop that to connect a lines via center to avoid crowding

baptiste Over a year ago

I was thinking of p + annotation_custom(gridExtra::arcTextGrob()), but it's incompatible with polar coordinates, sadly. Maybe it should only be a warning.

baptiste Over a year ago

there could be a geom_arc with associated stat, which would work like geom_segment, but never munched, and would connect pairs of observations with a curveGrob

Andy W Over a year ago

Cool example, reminds me of circos diagrams. My command of geometry isn't strong enough though to give suggestions about how to map such arcs in polar coordinates. Here is another example in d3.js as well. Edit - I see Paolo's answer gives a very similar example as well.

shNIL Over a year ago

Unfortunately circos is perl and no similar quality r package I could find in R.

Paolo · Accepted Answer · 2012-07-25 09:59:18Z

8

Take a look at ggbio. This package extends ggplot2 and the grammar of graphics to sequence data (bioinformatics) but they seems to solve your problem (for example see here). Taking a look at the source code for the package may direct you to a more generic solution.

answered Jul 25, 2012 at 9:59

Paolo

2,8251 gold badge21 silver badges23 bronze badges

Comments

BSMP · Accepted Answer · 2018-12-21 13:39:00Z

Add the ymin etc to fill d

fd <- data.frame(filld, ymin = 4, y =6, ymax = 4.5)

Then use geom_rect with the column label as the fill and colour aesthetics

 ggplot(fd, aes(x=start,xmin = start, xmax = end, y=y)) +  
   geom_rect(aes(ymin=ymin, ymax=ymax, fill = label )) + 
   ylim (c(0,6)) + 
   coord_polar()

Add the lines:

## calculate the midpoint for each segment
fd$xmid <- apply(fd[,1:2],1,mean)
## get the replicate id for the labels (what occurence is this)
library(plyr)
library(reshape2)
fd1 <- ddply(fd, .(label), mutate, id = 1:length(xmid))
## reshape to wide, subsetting only with more than one rep
.lines <- na.omit(dcast(fd1, label~id, value.var = 'xmid'))
## add a mid point between the mid points
.lines$mid <- apply(.lines[,2:3],1,mean)
## reshape to long, and add some y values 
ld <- data.frame(arrange(melt(.lines,value.name = 'x'), label, x), y = rep(c(4,3,4),2))

ggplot(fd) +  
  geom_rect(aes(x=start,xmin = start, xmax = end, y=y,ymin=ymin, ymax=ymax, fill = label )) + 
  ylim (c(0,6)) + 
  coord_polar() + geom_path(data = ld, aes(x=x,y=y,colour = label))

The lines are ugly, but there!

Collectives™ on Stack Overflow

Circular plot in ggplot2 with line segments connected in r

3 Answers 3

5 Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

5 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related