[Bioc-devel] Heatplus Heatmap Fails for Hidden Dendrogram and Labels

Hi Dario,

thanks for the feedback. Let's take this one by one:

plot(regHeatmap(expressionMatrix))

shows a heatmap.

When I add an extra parameter

plot(regHeatmap(expressionMatrix), list(status = "hidden"))

I get the error

Error in plot.new() : outer margins too large (figure region too small)
In addition: Warning message:
In pad1.rm.cm(widths, cm.widths, len = num.cols) :
  NAs introduced by coercion

It's strange that the figure region is too small, because it's supposed to be plotting less than for the first example, which worked.

The plot.new() error is typical for a too small output device - increase
the window size or the width/height of the file you are plotting to, and
it should go away. This may happen even if the plot causing the error is
technically "smaller" than one which does not - for each plot, the
device is split into multiple parts (using layout()), and the error will
occur if just one of the parts (say for the legend) is too small.

Also, the documentation and the vignette are inconsistent about the values that can be used. In the vignette is the example has status="hide" but the function's documentation writes "hidden means re-rorder".

Yeah, that's a bit of sloppy programming on my side: any value that is
not "yes" or "no" will be interpreted as "hidden", so both status="hide"
and status="hidden" will work the same way.

Providing row and column labels also fails.

str(expressionMatrix)
 num [1:45, 1:18] 18.56 46.65 4.52 52.12 0.26 ...

plot(regHeatmap(expressionMatrix, labels = list(labels=list(Row = 1:45, Col = 1:18), side = list(Row = 4, Col = 1))))
Error in mmar[x$labels$Row$side] = x$labels$Row$nrow : 
  invalid subscript type 'list'

You have to switch the "labels" and Row/Col in the command:

## Example setting labels
require(Heatplus)
example(regHeatmap) ## generate some data
map1 = regHeatmap(mat, labels=list(Row=list(labels=1:20),
Col=list(labels=1:5)))
plot(map1)


HTH, alexander

Alexander Ploner
Medical Epidemiology & Biostatistics
Karolinsksa Institutet
http://ki.se/meb

Thanks. That solves most of the problems.

When I resize the plot window of Rstudio to any size, it still gives an error. Can you reproduce this with random data ? Here is an example that creates a large PDF, but still causes the error.

values <- matrix(rnorm(45*18), nrow = 45)
pdf("tmp.pdf", h=40, w=40)
plot(regHeatmap(values), list(status="hidden"))
dev.off()