I believe this to be a bug with Avogadro:
(Please describe your bug or issue)
Expected Behavior
optimize geometry
Actual Behavior
Appimage deletes the structure, windows optimizes geometry
Steps to Reproduce:
This is a particular molecule glycyrrhizin. I loaded it from Import-> download by name. both windows and appimage loaded it ok. I then asked it to optimize geometry and the Appimage apparently started but wound up deleting the structure and going back to a blank screen. Windows version optimized the geometry and displayed the result.
I should note that the downloaded structure is entirely planar (and clearly should not be) so the problem may be with the structure downloaded and not the program.
John
Please upload files if appropriate here (or via file-sharing service like Dropbox or Pastebin)
The problem is that the structure “exploded.” I’m kind of surprised the Windows version produced a reasonable 3D structure.
When you get a 2D file like this, the initial coordinates are designed for display (e.g, ChemDraw) and not actually 3D realistic geometries. There are nearly-overlapping hydrogen atoms, etc.
In the current development version, if the default “download by name” returns a 2D geometry, Avogadro attempts to get one from PubChem. In this case, the molecule is too big, so that fails and returns an error. It’s not ideal since you’d like the molecule, but I think it’s better than returning a 2D geometry that causes further problems (e.g., is the stereochemistry correct?)
BTW, I took the SMILES from PubChem and generated this. I’ll add the script to the RDKit plugin later today e.g., “Insert => SMILES (RDKit)” - it generates 25 initial conformers, does a quick optimize and picks the best one.
I’d suggest checking the stereochemistry, but a quick glance looks okay.
Thank you for your reply. I was confused to why it would dump a planar structure on me. When it exploded, I went windows to compare and it succeeded. I am going to check what it produced with the stereochemistry to see if it was correct. I would guess it probably isnt, but I will check. Again, thanks for helping understand this.
Just to be clear - that was the stereo when Windows optimized a 2D structure?
I ask because that’s basically a coin-flip, and the SDF I uploaded was created with RDKit which definitely checks stereochemistry when creating 3D geometries.
The latest RDKit plugin should now offer a Build ⇒ Insert ⇒SMILES (RDKit)… command that uses RDKit to generate the 3D coordinates. That’s basically what I did for glycyrrhizin.
glycyrrhizin-licorice.cjson (16.4 KB) Here is the file that windows came up with. I didn’t check all the centers, but the first 3 or 4 I checked were good, then I hit a couple that were not.. I will give the RDkit a try. Thanks,
I looked at both ends and the sdf does not look correct. On the top ring the two methyl groups are shown here as trans (E), but on the sdf they look cis (z). On the bottom ring the carboxyl and adjacent OH are cis in the sdf and trans in the NIH molecule. I didn’t go any further as I wasn’t able to superimpose either of the top or bottom rings to match the rings in the NIH molecule. I didn’t go any further but they don’t appear to be the same. It would really be good to be able to have the computer label the centers as R or S. Been a long time since I have done this stuff and I am never quite certain I have it all right on a molecule this complex.
and I am already saying I am wrong on the bottom ring. I was looking at the 2nd ring and the 3d image was folded so I thought I was looking at the end ring. I think the comments on the top ring methyls still stand.
I am looking at it again today. I found that the png version keeps slipping on the screen. It moved the bottom ring off the page, and I wound up thinking the 2nd ring was the first as it was on the bottom. I will get back to you on this.
These are definitely diastereomers, not the same. I have reviewed the bottom ring and all of the centers are the opposite of the NIH picture except the ether linkage to the next ring. The top ring remains as I said at first the 1,4 methyl groups are on the same side of the ring not opposite. So this thing is neither the same nor the mirror image of the NIH picture. (assuming it is correct)
