dittoHeatmap with pale heat map colors

[kmakat]

Hi,

Thanks for that great package.

Just started to explore it. When I use dittoHeatmap on Seurat objects it runs without errors and produces the expected output in terms of annotation etc. but the heat map colors are pale, like there is a some layer on top of it. Changing the backend to ComplexHeatmap does not help.

Extracting the data and using ComplexHeatmap directly, works.

Maybe this a macOS/Quartz related issue?

Thanks, Kemal

M2 Mac
R Under development (unstable) (2023-10-30 r85440)
Platform: aarch64-apple-darwin20
Running under: macOS Sonoma 14.3.1
dittoSeq_1.15.3 

[dtm2451]

Hmmm very odd and I'm tempted to say yes, it might be a macOS/Quartz related issue but only because I don't have another explanation.

I think of dittoHeatmap() as "just" a wrapper that helps build pheatmap() inputs for the user, then shuttles them over. There's nothing internal to dittoHeatmap() that adjusts your given heatmap.colors (the pheatmap() color equivalent) or adds a layer to cause this paleness. Thus, the issue must come later on.

I have noticed in the past, when mucking with dittoHeatmap outputs in Illustrator, that there was a white background and the heatmap elements weren't completely opaque. But it is very odd to me that the difference is so profound for you as it has always been small enough for me that I hadn't even noticed until I moved to Illustrator!

After poking at the ComplexHeatmap::pheatmap() code, as you say you've seen the same with complex = TRUE, I'm still not sure at all where the effect might come from. There's nothing in that function itself which seems to give an opacity adjustment, unless there's something later on that derives from this line... If I use ComplexHeatmap::draw(<ditto_out>, background="black") the heatmap itself still looks the same as when background="white" for me. Is that the same for you on your M2 Mac?

[j-andrews7]

This is due to the scales being much larger in the first plot. Providing the breaks directly will fix it.