R help - Apr 2012 - extend data frame for plotting heat map in ggplot2

Hi all!

I want to generate a heat map from an all-vs-all comparison. I have the 
data, already scaled to 0-1. However, I have the values only for the 
comparisons in one way, and not for the comparisons between the same 
group (which are always 1), i.e. I have half the matrix and am missing 
the other half and the diagonal.
What is a good way to get it into a form that ggplot2 can use for the 
heat map?

This is an example of the data I have:

A	B	value	
T1	T2	0.347
T1	T3	0.669
T2	T3	0.214

I assume the following is what I need for ggplot (or maybe I don't, if 
ggplot can somehow generate it?):

A	B	value	
T1	T2	0.347
T1	T3	0.669
T2	T3	0.214
T2	T1	0.347
T3	T1	0.669
T3	T2	0.214
T1	T1	1
T2	T2	1
T3	T3	1

Then I would run

sorted<-data[order(data$A, data$B), ]

ggplot(sorted, aes(A, B)) +
   geom_tile(aes(fill = value), colour = "white") +
   scale_fill_gradient(low = "black", high = "red") +


I have solved this, but in (what I assume is) a really bad way involving 
for loops. There must be a better way to get form the first data frame 
above to the second!


Cheers,

Till

-- 
Till Bayer
4700 King Abdullah University for Science and Technology
Building 2, Room 2227
Thuwal 23955-6900
Saudi Arabia
Phone: +96628082373

On Sun, Apr 1, 2012 at 9:16 AM, Till Bayer <till.bayer at kaust.edu.sa>
wrote:
> Hi all!
>
> I want to generate a heat map from an all-vs-all comparison. I have the
> data, already scaled to 0-1. However, I have the values only for the
> comparisons in one way, and not for the comparisons between the same group
> (which are always 1), i.e. I have half the matrix and am missing the other
> half and the diagonal.
> What is a good way to get it into a form that ggplot2 can use for the heat
> map?
>
> This is an example of the data I have:
>
> A ? ? ? B ? ? ? value
> T1 ? ? ?T2 ? ? ?0.347
> T1 ? ? ?T3 ? ? ?0.669
> T2 ? ? ?T3 ? ? ?0.214
>
> I assume the following is what I need for ggplot (or maybe I don't, if
> ggplot can somehow generate it?):
>
> A ? ? ? B ? ? ? value
> T1 ? ? ?T2 ? ? ?0.347
> T1 ? ? ?T3 ? ? ?0.669
> T2 ? ? ?T3 ? ? ?0.214
> T2 ? ? ?T1 ? ? ?0.347
> T3 ? ? ?T1 ? ? ?0.669
> T3 ? ? ?T2 ? ? ?0.214
> T1 ? ? ?T1 ? ? ?1
> T2 ? ? ?T2 ? ? ?1
> T3 ? ? ?T3 ? ? ?1
You can usually do something like:

df <- data.frame(A = 1:2, B = 3:4, value = runif(2))
all <- expand.grid(unique(df[c("A", "B")]))

merge(all, df, all = T)

Hadley

-- 
Assistant Professor / Dobelman Family Junior Chair
Department of Statistics / Rice University
http://had.co.nz/