heatmap.sim colors switched?

Hi!

I just reanalyzed a dataset that I had worked on with an older mothur version, and it seems the colors of the heatmaps are swiched. Red is less similar, and black more similar, the opposite of what my old files show. I can’t really think of a way this could be user error, or is it?


Cheers,

Till

Thanks for bringing this to our attention. This is a bug in version 1.20 and 1.21. In version 1.20, we changed the output of mothur’s calculators to be distance instead of similarity values. This change reversed the colors outputted by the heatmap.sim command. This will be fixed in our next release.

Hi again!

It’s been a while since I used heatmaps, but was this bug ever fixed (or fixed and introduced again?) The current verison (1.24.1) still produces heatmaps that are colored the opposite of what would be expected from the summary.shared command values.

Cheers,

Till

Hi,
I think this has been fixed. Looking at the example from the Schloss SOP:

summary.shared(shared=final.an.shared, calc=jest, distance=t) creates this distance file.

11
F003D000
F003D002 0.720683
F003D004 0.747450 0.555709
F003D006 0.688352 0.701856 0.683551
F003D008 0.728822 0.698880 0.721511 0.675200
F003D142 0.763922 0.733082 0.735997 0.734071 0.718849
F003D144 0.684296 0.719047 0.772515 0.706863 0.696744 0.624007
F003D146 0.723000 0.782287 0.692952 0.697834 0.741914 0.679278 0.597831
F003D148 0.768403 0.789137 0.771942 0.739037 0.798618 0.602999 0.654353 0.669791
F003D150 0.802176 0.813291 0.810950 0.776612 0.806281 0.739575 0.731990 0.699632 0.775556
MOCK.GQY1XT001 0.994830 0.994399 0.992751 0.993127 0.992000 0.996775 0.996311 0.996780 0.992233 0.996173

so the similarities would be:

11
F003D000
F003D002 0.279317
F003D004 0.252550 0.444291
F003D006 0.311648 0.298144 0.316449
F003D008 0.271178 0.301120 0.278489 0.324800
F003D142 0.236078 0.266918 0.264003 0.265929 0.281151
F003D144 0.315704 0.280953 0.227485 0.293137 0.303256 0.375993
F003D146 0.277000 0.217713 0.307048 0.302166 0.258086 0.320722 0.402169
F003D148 0.231597 0.210863 0.228058 0.260963 0.201382 0.397001 0.345647 0.330209
F003D150 0.197824 0.186709 0.189050 0.223388 0.193719 0.260425 0.268010 0.300368 0.224444
MOCK.GQY1XT001 0.005170 0.005601 0.007249 0.006873 0.008000 0.003225 0.003689 0.003220 0.007767 0.003827

heatmap.sim(shared=final.an.shared, calc=jest)

Looking at the svg image created the MOCK.GQY1XT001 is almost black, indicating low similarity.

Are you seeing different results with your files?

Kindly,
Sarah

I don’t know if this has been discussed/fixed already, but I do seem to have the same problem.

That, or I’m interpreting the heatmaps or diversity indices the wrong way.


When I create a heatmap based on theta-Yue-Clayton looking for similarities between 6 samples, the similarity between e.g. sample F6 and F9 is shown as bright red, being between 0.8 and 1.0

When I perform the summary.shared command, using the same input file and the same calculator, the similarity between F6 and F9 turns out to be 0.04.

So, what does the heatmap.sim show? Distances or similarities?

ps: I have been using version 1.24.1 so far.
edit: I just updated to 1.27.0 and the output remains the same

The heatmap command displays similarities, but the summary.shared command outputs distances. A distance of 0.04 would mean a similarity of 0.96, which should be bright red.