We recently posted our first âcommunity surveyâ to gather feedback. Over a week,
we received 66 responses. The survey was highlighted on Twitter and on the
avogadro mailing lists.
This is a companion discussion topic for the original entry at https://avogadro.cc/news/community-survey-results/
The results are interesting - in part because there are definitely users unaware of features already in the software.
Hopefully some of those giving up 5-10 minutes for a survey can be convinced to come to the forum or submit bug reports.
A post was split to a new topic: Duplicating atom
I think I was one of those asking for a more complete conformer search feature. From what I can tell, conformer search only rotates unbounded sp3-sp3 bonds. So a conformer search on a substituted seven membered ring isn’t possible. Sp2-Sp3 bonds are also not included. For example, propanal doesn’t give you any searching options although there are two low energy conformations. So, unless I’m missing something, conformer searching is currently pretty limited.
I’m not much of a coder, but it seems that it might not be too difficult to do external coordinate Monte Carlo conformer searches as described by Saunders and Still (J. Am. Chem. Soc. 1990, 112, 1419-1427). It’s mostly file manipulation and random number generation. I think the hard parts, minimization and structure comparison, are already done.
I don’t like to tell people to do things I’m not willing to do myself, so maybe I should rediscover coding and do it myself. Short of that I’d be willing to help test if someone else wanted to give it a try.
All said, I think you’ve done a great job with Avogadro. Keep up the good work.
Thanks for the reply! At the moment, Avogadro uses Open Babel for conformer searching. The current shipping versions of OB don’t do ring conformers because they’re a bit harder to enumerate. (The development version has something, but it’s not ideal.)
I’d certainly be willing to help you work on any type of ring-conformer algorithm as it’s something that interests me.
There are two broader efforts at work that I mentioned in the write-up.
- We’re working with integrating Avogadro v2 and other packages, notably RDKit (http://rdkit.org/). The distance-geometry and ETKDG methods in RDKit implicitly sample ring conformers. I’d be happy to point you at some scripts that use RDKit, and there will definitely be an “RDKit conformer” plugin sometime in 2018.
- My group is working on more sophisticated conformer searching code that should involve better ways to sample correlated torsions involved in rings. That’s a research project, but I hope we’ll have some initial code out in the next 12 months.
I’ll take a look at that Still paper soon and definitely drop me an e-mail if you’re interested in helping implement it.
The interesting thing about the Saunders/Still paper is that the more complicated ring breaking-reforming algorithms that Still developed were actually outperformed by the much simpler “just kick the atoms around” method developed by Saunders. Saunders’ method kind of obviates the need to do correlated torsions. You just whack the molecule around and test that the chirality stays intact. The Saunders’ Monte Carlo approach became the default method used in MacroModel, which is now out of my price range. It didn’t get more expensive, I just got poorer.
Still published a later paper(J. Am. Chem. Soc. 1995,117, 8211-8219) explaining in more detail how they further implemented the Monte Carlo approach.