plotDiffHeatmap: avoid failure when only one cluster remains

[SamGG]

IIUC, when there is only one cluster left, as.matrix() results in a one column matrix instead of one row matrix.

CATALYST/R/plotDiffHeatmap.R

Lines 243 to 250 in f3e294e

	da = {
	ns <- table(x$cluster_id, x$sample_id)
	fq <- prop.table(ns, 2)
	fq <- fq[top$cluster_id, ]
	y <- as.matrix(unclass(fq))
	if (normalize) y <- .z_normalize(asin(sqrt(y)))
	Heatmap(
	matrix = y,

I propose to transpose y in such a case.
if (nrow(top) == 1) y <- t(y)
In my hands, it works and permits to display a single significant cluster.

[SamGG]

In fact, the following is even simpler:
fq <- fq[top$cluster_id, , drop = FALSE]

[HelenaLC]

Sounds good. Thanks again for catching this. Edgy cases aren't my strong suit to think of ;) - will do along with issue #381.

[SamGG]

Thanks. Edgy cases are very difficult to anticipate.
This question is independent from #381.

[HelenaLC]

Yes, absolutely, I only meant that I will try and resolve both with one update/version bump.