HeatmapAnnotation not working within mapply call

[flde]

Hi,

Many thanks for your great work and tutorials. However, I encountered a problem I would need some help with.

When I call HeatmapAnnotation(foo1 = runif(10)) in my script everything works perfectly. But when I call it embedded in a function with mapply it breaks.
plot_FUN <- function(x, y) {HeatmapAnnotation(foo1 = runif(10))}
mapply(plot_FUN, list(a=1), list(a=1), SIMPLIFY = FALSE)

Error: It seems you are putting only one argument to the function. If it is a simple vector annotation or a function annotation (e.g. anno_*()), specify it as HeatmapAnnotation(name = value). If it is a data frame annotation, specify it as HeatmapAnnotation(df = value)

I thought it is a mapply problem but for example ComplexHeatmap::pheatmap works with mapply. And it works with lapply. Any ideas?

lapply(list(a=1), plot_FUN)

I work on JupyterLab with R 4.1.0 kernel and ComplexHeatmap 2.8.0.

All the best,
Florian

[jokergoo]

I think I only tested HeatmapAnnotation() with lapply() and do.call() while not with mapply(). I will have a check.

[flde]

Many thanks, I think it is always possible to adapt lappyl() or do.cal() so I mark as answered.

Maybe one comment. If I use as.grob(expression(plot(object))) within a lapply() function, as.grob will not find object. I have to assign the objec to the global environment first within the function call by object <<- object.

[jokergoo]

I have never used it like that.

Why don't you use grid.grabExpr(plot(object))?

[jokergoo]

What is your purpose of doing the following two?

lapply(list(a=1), plot_FUN)

and

plot_FUN <- function(x, y) {HeatmapAnnotation(foo1 = runif(10))}
mapply(plot_FUN, list(a=1), list(a=1), SIMPLIFY = FALSE)

Do you want to automatically generate a list of annotations? Why don't you use do.call()?

do.call("HeatmapAnnotation", list(a = 1:10, b = letters[1:10]))

[pormr]

I've encountered the same problem recently. After some sort of investigations, I found out that purrr::map2 can be used as a drop-in replacement of mapply in some cases. For more than two arguments, you can also use foreach function from foreach package.

[jokergoo]

I think the problem is I want to automatically identify what users have specified in HeatmapAnnotation() function. I can correctly capture everything when HeatmapAnntation() is directly used, but I haven't considered complex use cases like with mapply() or maybe sapply() also.

[jokergoo]

These scenarios I have already tested:

HeatmapAnnotation(a = 1:10)
rowAnnotation(a = 1:10)
columnAnnotation(a = 1:10)
do.call(HeatmapAnnotation, list(a = 1:10))
do.call(rowAnnotation, list(a = 1:10))
do.call(columnAnnotation, list(a = 1:10))
do.call("HeatmapAnnotation", list(a = 1:10))
do.call("rowAnnotation", list(a = 1:10))
do.call("columnAnnotation", list(a = 1:10))

But I haven't tested these scenarios:

f = function() HeatmapAnnotation(a = 1:10)
f()
f = function() rowAnnotation(a = 1:10)
f()
f = function() columnAnnotation(a = 1:10)
f()

sapply(1, function(x) HeatmapAnnotation(a = 1:10))
sapply(1, function(x) rowAnnotation(a = 1:10))
sapply(1, function(x) columnAnnotation(a = 1:10))

mapply(function(x) HeatmapAnnotation(a = 1:10), list(1), list(1))
mapply(function(x) rowAnnotation(a = 1:10), list(1), list(1))
mapply(function(x) columnAnnotation(a = 1:10), list(1), list(1))

I will have a close look.

[jokergoo]

@flde @pormr In the previous versions, I made validations on annotations too complicated. Now I simplified them and it is working now. PS: I still need to make more tests to ensure there is no more error, but at least your example can be run successfully.

> plot_FUN <- function(x, y) {HeatmapAnnotation(foo1 = runif(10))}
> mapply(plot_FUN, list(a=1), list(a=1), SIMPLIFY = FALSE)
$a
A HeatmapAnnotation object with 1 annotation
  name: heatmap_annotation_2
  position: column
  items: 10
  width: 1npc
  height: 5mm
  this object is subsettable
  9.11106666666666mm extension on the right

 name   annotation_type color_mapping height
 foo1 continuous vector        random    5mm