How to create color range for RPKM/FPKM using R?

Question

[Whoknows]

Hi every body

I want to create a color range for FPKM/RPKM data using R or another software if it possible, but i don't know how to do it.

For better understanding please take a look at the photo.Thanks a lot

[Sean Davis]

I might suggest using a density plot of values instead. It will be much more informative, I suspect.

[Devon Ryan]

Have you read the documentation for scale_gradient()?

[Whoknows]

yes i've read this kind of scale_gradient() diagram but as you see in the chart my values are discrete,

please note that i have more than 10000 FPKM/RPKM values which will be show in the diagram,

anyway thanks again,

[Devon Ryan]

Err, your FPKM/RPKM values are continuous, that's why you can simply color by value and use a gradient scale, as in that image.

[Devon Ryan]

BTW, you probably want to color on log2 level.

[Whoknows]

yes of course log2, but how !!??

[Devon Ryan]

The simplest way would be to just graph log2(RPKM) to begin with. Alternatively, just add that as a column to the dataframe or just perhaps even use something like colour=log2(RPKM) as an argument.

[Michael]

What do you want to use the color gradient for, a heatmap?

Answer 1

[Whoknows]

I offer this way if anybody expert in R please develop it

x <- data.frame(FPKM = rnorm(1000))
x$Color <- cut(x$FPKM, 256)
library(gplot)
library(ggplot2)
x$Color2 <- redgreen(256)[as.numeric(x$Color)]
ggplot(x, aes(FPKM,1,fill=Color2))+geom_bar(stat="identity")+coord_flip()
// Or use below after log2
ggplot(y, aes(FPKM,ymin=-1,ymax=1,color=Color))+
geom_linerange()+coord_flip()+theme(legend.position="none")+scale_x_log10()

[Jason]

Is library(gplot) supposed to be library(gplots)? I tried the graph, and it seems like kind of an odd way to represent the data. Also I get the following error when using the code

> ggplot(y, aes(FPKM,ymin=-1,ymax=1,color=Color))+
+ geom_linerange()+coord_flip()+theme(legend.position="none")+scale_x_log10()
Error: ggplot2 doesn't know how to deal with data of class numeric

Did you copy the code exactly? I agree with the previous poster that said a histogram would be much more informative. They are fairly simple to make, just use the hist() function.

hist(x[,1])

Alternatively, you could plot the data as a line or xyplot where the cumulative percent (or actual number) of genes is on the y-axis and RPKM value (log2) is on the x-axis.