FAQ QM DMD

Issues and Questions for QM/DMD

NOTE: Before checking this page for errors with your QM/DMD simulation, check the Turbomole and DMD FAQ pages. As QM/DMD is an amalgam of Turbomole and DMD, it holds many errors in common with these two programs.

1. My ligand adopts unphysical conformations or stops binding to my protein during my simulation

DMD does not generate good parameters for many types of ligands. The ligand could be your substrate, a cofactor, etc. It struggles with fused ring systems and highly charged functional groups (such as phosphates) in particular. There is no direct fix to this issue as we currently have no good way to consistently modify DMD force field terms for unusual molecules.

The best fix involves freezing the most problematic regions of your ligand. If your ligand can be incorporated into your QM region this entails careful construction of your QM region. Include the problematic groups directly in your QM region and freeze them in DMD. Make sure to include in your QM region residues that hydrogen bond to the problematic groups of the ligand. For some ligands, it may also be necessary to include in your QM region non-polar groups that apply steric constraints to the problematic groups - so that the QM geometry optimization does not move your ligand to a clashing structure with the overall protein geometry.

If your ligand can not be incorporated into your QM region, the solution is less appealing. You can still freeze the problematic groups in DMD, but this will adversely impact your simulation if (a) the ligand is large (such as a peptide of more than a couple residues), (b) the flexibility of the ligand is important for the function of your protein, or (c) the ligand is particularly far from your QM region (this severely reduces the conformational flexibility of the protein as at least part of your QM region is also frozen - two frozen regions far from each other means that the protein can not effectively move around them). In any of these cases, it may be better to run simulations that exclude the ligand.

2. I got a SCF failed to converge error

QM/DMD simulations involve single point and geometry optimization calculations on often large QM regions. A failure of SCF convergence means that Turbomole could not obtain a stable electronic structure for your QM region. To fix this problem, first check the geometry of your QM region to make sure it looks physical. Next, check the overall charge of the system and ensure that the correct charge is assigned in the phdinput.json file. Finally, if nothing else seems off increase the number of SCF cycles allowed in the phdinput.json file - found as the "iter" keyword under the "scf" section. Some systems, particularly large ones, can require upwards of 3000 SCF cycles to converge. Some systems are simply too large; do not expect a QM region of over 160 atoms to consistently and readily converge.