Is it possible with openbabel to generate random 3D geometries (i.e. conformers) of a molecule without taking energy into account? Ideally with the property of being truly random, so that after generating conformers like this for a sufficient amount of time, a good coverage of the conformational space is obtained. Because the --conformer options of openbabel takes energy into account, but I want to input the geometries into a qc program that then calculates the energies for the conformers.
$\begingroup$
$\endgroup$
-
1$\begingroup$ I would generate a normal coordination file and write a little script that twists desired bond-angles by different amounts $\endgroup$ – Fl.pf. Jul 6 '17 at 9:18
-
3$\begingroup$ If you create actual random 3D geometries I expect 99.9% of those being absolute garbage and they won't have anything to do with different confomers. There are techniques which allow for confomer search, for example simulated annealing. $\endgroup$ – DSVA Jul 6 '17 at 9:27
-
$\begingroup$ ^ this. I don't know what exactly you want to achieve, but maybe some kind of MD simulation could help you get enough snapshots? I mean you have something in the scale of $n_{steps} \cdot N_{atoms}^3$ calculations to do if you want a random, statistical approach to all angles. I don't know, but my boss would ask me if I was crazy if I would start 10k calculations of which 9990 would be garbage... $\endgroup$ – Fl.pf. Jul 6 '17 at 10:49
-
$\begingroup$ Well, the thing with the openbabel conformer generator is that it is often not expolaritive enough (sometimes even only a single conformer is returned). So my idea was to generate random conformers and optimize them afterwards to have a more robust method. $\endgroup$ – logical x 2 Jul 6 '17 at 16:12
-
$\begingroup$ 99.9% no problem: then I just generate 10000 structures and get 10 conformers that are actually different. because openbabel just gives me basically always the same conformer. $\endgroup$ – logical x 2 Jul 6 '17 at 16:16
$\begingroup$
$\endgroup$
Random conformer search generates random conformers
Open Babel, like many conformer sampling tools, has a set of common dihedral angles to sample. This minimizes the search space, e.g., only 3-4 likely cases for each rotatable bond.
The RandomConformerSearch method indeed generates conformers using a random sampling of dihedral angles. It then (optionally) runs a force field for some number of steps to minimize any steric clashes. No energies are considered.
If you're finding only one conformer output, that's because you're not using the --writeconformers flag:
http://open-babel.readthedocs.io/en/latest/3DStructureGen/multipleconformers.html
answered Oct 12 '17 at 16:48
-
2$\begingroup$ PS - I only recently saw this question. There is an Open Babel support forum - I'm not coming to Chem.SE as often as I'd like. $\endgroup$ – Geoff Hutchison Oct 12 '17 at 16:49
-
$\begingroup$ OK, I will keep the open babel support forum in mind. But I tried several combinations of input flags and always got a single conformer. If I have time on my hands, I will try to make a minimal example of the problem I ran into and post it into the open babel forum. By the way, Avogadro and OpenBabel are great, keep up the nice work : ) Anything else in planning? $\endgroup$ – logical x 2 Oct 12 '17 at 20:53
-
$\begingroup$ Doesn't matter what combination of input flags - if you don't use
--writeconformersyou'll only get the lowest energy one. Alternatively, write a Python script. :-) $\endgroup$ – Geoff Hutchison Oct 12 '17 at 21:28
