As noted in its homepage, JME is deprecated. Unfortunately, the proposed replacement, JSME, does not seem to be readily interfaced with Mathematica (but this may be due to my ignorance of how to invoke and use JavaScript objects in Mathematica).

Thus,

1. Are there other Java-based solutions for chemical drawing that are a. small; and b. easily interfaced via JLink?

or

2. Is there a straightforward way to get JSME to work with Mathematica?

I am planning to eventually have MoleculeViewer support the plotting of various molecule-related surfaces, like the solvent-accessible (Lee-Richards) and solvent-excluded (Connolly) surfaces, as well as electrostatic potential surfaces and molecular orbitals down the road. These represent a fair amount of nontrivial effort (ContourPlot3D[] is often too slow for plotting these), and the algorithms for implementing these efficiently are not that straightforward to do in Mathematica.

Version 3 of MoleculeViewer is coming out Soon™. Among many new features that have been added, there is now a new style of depicting multiple bonds. As a preview of this upcoming functionality, here is a side-by-side comparison of a depiction of cimetidine using the old and new styles:

The new style preserves the tetrahedral angle around atom centers, just like the old parabolic depiction of multiple bonds, but looks "tighter".

Additionally, the new style also seems to look better when used along with tube and wireframe depictions:

(At the very least, both look better than the currently-used depictions by MoleculePlot3D in version 12 of Mathematica.)

Now, for the issue: I am not sure what to call this new style. Currently, I use "Classic" to refer to the old parabolic bonds used by MoleculeViewer, and "Taut" to refer to these new style bonds. I have not yet settled on whether this should be an option setting or something else.

I have been thinking if I should implement an aromaticity indicator in MoleculeViewer. Something like this:

The gnarly part is checking if Hückel's rule is satisfied in the component rings, but there are a number of gotchas in that that coding it would be a nontrivial effort.

Would this be worthwhile?

Currently, MoleculeViewer is using ChemSpider, PubChem, NCI CACTUS, and the built-in ChemicalData[] as a source of 3D coordinates of molecules. Are there other widely-used chemical databases that a. have 3D coordinates of molecules (supplied as e.g. MDL MOL files); and b. is accessible through an API? I might be able to add support for these as additional sources for MoleculeViewer.