Overlayed Circular multiple barcharts in R

Question

The following is my data:

chr <- rep (1:4, each = 200) 

position <- c(1:200, 1:200, 1:200, 1:200)

v1bar <- rnorm(800, 10, 2)

v2bar <- rnorm(800, 10, 2)

v3bar <- rnorm(800, 10, 2)

mydata <- data.frame (chr, position, v1bar, v2bar, v3bar)

I want to create multiple circular bar charts with the x value = pos and y value = v1bar, v2bar, v3bar (all three will be in successive circles). Each circle is divided to chr. Thus each circle has "pie slice". I am not sure what such graph is known as and would it be possible to develop one. The following just rough sketch to illustrate my idea.

enter image description here

Edits: My assuption is someway similar to the following circos figure.

http://circos.ca/tutorials/images/zoom/8.8

EDITS:

In response to the following answer to David, this what I have imagined - chr not seperate circe but classifying as different slice (such as in circos figure) enter image description here

rbloggers had a similar graph (histogram) discussed earlier this week CLICK HERE — Tyler Rinker
What specifically do you want to include from the circos figure that my answer, for example, doesn't do? — David Robinson
Your figure all features except (1) chr are different circle (instead of var) - which obivously can change easily (2) Chr are seperate slice of pie in addition to var in different circles - circos figure has this property — jon
See my most recent addition- I think it's just what you're looking for. — David Robinson

David Robinson David Robinson · Accepted Answer · 2012-02-23T17:32:51

Since you are presenting chromosomes in a circular way, try the tools provided by the ecolitk package provided by Bioconductor, which includes tools for plotting various kinds of shapes on circular chromosomes.

ETA: Here is an example of its use to create a circular bar plot, though it only scratches the surface of the things you can do with it.

circular plot

library(ecolitk)

plot.new()
plot.window(c(-5, 5), c(-5, 5))

plot.chrom = function(data, chromlength, radius=1,
                        width=chromlength/length(data), ...) {
    linesCircle(radius, ...)
    starts = seq(1, chromlength - width, width)

    scale = .5 / max(abs(data))
    for (i in 1:length(starts)) {
        polygonChrom(starts[i], starts[i]+width, chromlength, radius,
                data[i] * scale + radius, ...)
    }
}

plot.chrom(rnorm(100, 10, 1), 10000, radius=1)
plot.chrom(rnorm(100, 10, 2), 10000, radius=2, col="blue")
plot.chrom(rnorm(100, 10, 5), 10000, radius=3, col="red")
plot.chrom(rnorm(100, 10, 10), 10000, radius=4, col="green")

legend("topright", legend=c("chr1", "chr2", "chr3", "chr4"),
       col=c("black", "blue", "red", "green"), lty=1)

ETA: Ah, now I understand what you meant about division of plots. In that case, this code should be what you're looking for. It uses the data you present (with one slight modification- I had to give the chromosome column a name so I could use ddply with it) and allows you to specify the spacing between chromosomes. While I haven't tested it in depth, varying things like the individual chromosome lengths and the mean and variance of the simulated data should work as you expect.

Circular plot divided

plot.multi.chrom = function(data, colors, spacing=50) {
    plot.new()
    plot.window(c(-5, 5), c(-5, 5))

    lengths = ddply(data, .(chr), function(x) max(x$position))
    nchrom = NROW(lengths)
    offsets = cumsum(c(0, lengths[, 2])) + cumsum(c(0, rep(spacing, nchrom)))
    tot.length = offsets[length(offsets)] + spacing

    scales = .75 / apply(abs(data[, 3:NCOL(data)]), 2, max)

    for (i in 1:NROW(data)) {
        for (j in 3:NCOL(data)) {
            start = offsets[data[i, 1]] + data[i, 2]
            polygonChrom(start, start + 1, tot.length,
                         j - 2, data[i, j] * scales[j - 2] + j - 2,
                         col=colors[j - 2])
        }
    }
}

chr <- rep (1:4, each = 200) 
position <- c(1:200, 1:200, 1:200, 1:200)
v1bar <- rnorm(800, 10, 2)
v2bar <- rnorm(800, 10, 2)
v3bar <- rnorm(800, 10, 2)
mydata <- data.frame(chr=chr, position, v1bar, v2bar, v3bar)

 require(plyr) 

plot.multi.chrom(mydata, colors=c("red", "black", "green"), spacing=50)

legend("topright", legend=c("V1", "V2", "V3"),
       col=c("red", "black", "green"), lty=1)

Overlayed Circular multiple barcharts in R

2 Answers