THe description says (and I remember it from my school lessons ^^) that you need an acceptor and a donator.
Look at the picture. All acceptor, 2 bonds.
Or is not the ball the defining thingy?
We need more information on the new visuals!
The underlying issue here is that there are cases where the same oxygen or nitrogen atom can act as both a hydrogen bond donor and a hydrogen bond acceptor (e.g. -OH and -NH2). So the tricky part is finding a concise way to state this…
To be even more exact where a donor and an acceptor are bonded to each other (as in the above groups), to reduce visual clutter, we simply show it as a purple acceptor/donor. It's like magnetism with a "middle class": blue attracts red, and purple is attracted to both and to itself.
The alternative to this abstraction would be to show a red and blue sphere mostly overlapping, but that just increases clutter and gains only pedantic precision, not actually useful precision.
On the other hand, the label seems off. That needs to be addressed.
Can you make the purple a bit more to white?
The color blind thingy. I can see it, but it is not very easy.
Alright, the label should be fixed, and the purple should be lighter in colorblind mode.
I found this program, and used it to simulate deuteranopia: http://colororacle.cartography.ch/. Hopefully this will help us come up with better colors in the future!
Marking as resolved since it has been released. Please reopen if there is still a problem.