Hi all,
Rather a simple request here. If I have a large molecule, say 55 to 70 atoms big, it is bound to have several energy minima. One example being, if this were a derivative of a sugar like glucose, there would be typically around two unique energy minima associated with a different orientation of the hydroxyl group.
I would like to collect as many converged minima as possible. Is there anyway to perturb the structure so it drives down to the nearest minimum?
A reason for doing this would be to collect as large a set of interatomic distances between certain atom e.g., if this were a zinc binding ligand, I would want to measure the distance between the zinc atom and the nitrogen atom at different rotations of a histidine which each are in an energy minima. This rotation may be the effect of a neighbour ligand (a further histidine).
Thanks
So what you want do to here is a conformational search to locate multiple minima on the potential energy surface. The task is not trivial and gets hard of the size of your molecule gets large - actually if the number of rotatable bonds gets large.
The way to it is to take certain gemetry and then change / rotate the rotatable bond by some given increments (let's say 60 degrees), saving the structure after each rotation. After you finish rotating the first bond, you start with the second one and so on to make every possible conformation. So for example for 2 rotatable bonds you will get 6*6 = 36 starting geometries, which you should optimize separately. Some of them will converge to the same conformations, but at the end you will get a set of different possible minima.
This is obviously problematic if you have a large number of rotatable bonds - 3 already pose a challenge since you get 6*6*6 = 216 starting conformations, and the more, the worse it gets. You can do 2 things here:
- build those 216 conformations and optimize them quickly in one of the semiempirical methods (I'd suggest PM6 or PM7). Then take the semiempirical minima (hopefully the number will be quite small) and reoptimize them in the desired QM method
- for larger number of rotatable bonds: there are programs that do conformational serach for really large ligands (e.g. Maestro/Macromodel from Schrodinger) with any number of rotatable bonds using smart algorithms and MM methods. If the systems is organic / bioorganic force-field based methods should give reasonable results and are very fast. Obviously you won't be able to scan the entire potential energy surface, just a fraction of it, but you should still get a number of low-energy conformations, which can be further optimized using more sophisticated methods. In these searches you can usually request a number of starting conformations and if you go with a realtively large number (I do usually at least 10,000 conformations for ligands with 50-70 atoms and 4-7 rotatable bonds) the entire search should take hours.
Zinc will pose problems, though, since it's likely not parametrized in these force fields. So you can try making the conformational search just for the organic parts and after you get a set of conformations you can add manually Zinc ion at different positions and reoptimize in QM.
Dear Anthony,
To add more possibilities:
1) With gaussian you can SCAN two parameters that you are interested in. To do it you must include the scan of two variable with the same amount of points (take into account that the total number of structures is the result of multiply the point, (i.e 10 point of a bond distance and 10 point of a dihedral angle will perform 100 structures optimizations). The results will be a 3D surface (P1*P2*Energy). To see how perform it, you can go to page 12 of my guide and, if you have any issue, send me a message.
2) Other program that could help you a lot with that is the Chimera program (free license). It is for bimolecular modelling but it is very intuitive and it can perform all the conformational studies and interactions.
I hope it helps you,
Joaquim Rius
https://www.cgl.ucsf.edu/chimera/
Technical Report Gaussian Guide. Section 1. Optimization and Frequency Calculations
There is no general procedure to find global minimum in quantum chemistry, Anthony. Moreover, different methods may give different numbers of minima and their order. It is nothing to do in quantum chemistry without chemical intuition.
I will rather do a potential energy scan search at low level of theory
I agree with all the above. In my opinion, freeze the known dihedral angles, which mimic the experimental structure(say the glucose structure obtained from the data base) and then relax others angles and do the scan using semi empirical or ab initio methods at a low level of theory. Once you get these minimum conformations, by relaxing the frozen dihedral angles and again do the minimizations using higher level of theory.
- Badges
- Science topic