Skip to content

QM DMD Options in phdinput.json

Jump to bottom
natjohnston edited this page Apr 27, 2022 · 3 revisions

phdinput.json Options

It is recommended that users are familiar with setting up jobs for TURBOMOLE and DMD seperately as QM/DMD is an amalgamation of the two. Hence, within the phdinput.json file are 2 sections that correspond directly to a dmdinput.json and definput.json (dmd params and qm params respectively). Users should consult the pages describing the options for these file when making adjustments. Only minor notes here will be used to emphasize how these sections differ within the QM/DMD engine.

The top portion of the default phdinput.json contains the QM/DMD parameters:

  • pdb file (string): the pdb file that will be used as the initial input.
  • Resubmit (bool): allows phd3 to resubmit the job on a queuing system if the time runs out. This should be set to false if you are not running on a super computer with a queuing system or have not set up the appropriate parameters in the phd_config.json file.
  • DMD CONVERGE (bool): determins if the engine will try to converge the DMD structure using the average trajectory potential energy. It is recommended that this be turned off.
  • MAX DMD STEPS (positive int): maximum number of DMD simulations to run to try and converge the structure. This parameter is ignored if DMD CONVERGE is false.
  • Max Iterations (positive int): maximum number of QM/DMD iterations to perform before stopping.
  • Cluster Energy (bool): determines if the minimum energy structure from each cluster should be used as a structure for the QM singlepoint calculations.
  • Cluster Centroid (bool): determines if the medoid (or centroid) structure from each cluster should be used as a structure for the QM singlepoint calculations.
  • Max Clusters (positive int): determines the maximum number of clusters to be allowed. A first pass with HDBSCAN allows for a flexible number of clusters. If the number of clusters surpases this value, then the single-linkage hierarchy cluster, with a fixed number of clusters, will be used instead resulting in the Max Clusters values specified here.
  • Equilibrate (dictionary): controls the equilibration of the dmd structure at the beginning of each iteration. The following parameters are available:
    • Equilibrate On (bool): determines if an equilibration will be performed prior to the first dmd simulation. If false, then it will ignore the rest of this section.
    • Initial Temperature (positive double): value to start the temperature at for the equilibration. This value will be gradually lowered to the Initial Temperature in the dmd params section so that the kinetic energy is the same as that when it starts the actual DMD simulations.
    • Time (positive int): Total number of DMD timesteps to perform the equilibration for.

Specifying the QM Region

The QM Chop section specifies to phd3 which atoms within the protein should be included in the QM region. By default all non-canonical amino acid residues (substrates and metals) are included but you can always exclude molecules in the Exclude Atoms section. Recall the identification of chain, residues, and atoms are described in the Input Files Primer.

The Residues (list of strings) specifies which residues to include in the active site. Hence

"Residues": ["A:117", "A:204"]

will include the sidechains of residue A:117 and A:204 in the active site (cut between the alpha and beta carbons and cap the beta-carbon with a hydrogen). It is also possible to select a range of residues and the backbone atoms connecting them using the following notation:

"Residues" : ["A:117N-A:120", "A:204", "A:205N-A:206C"]

Now we are selecting the residues A:117, A:118, A:119, and A:120 as well as the backbone for those atoms. The N and C specifiers at the end of the residue identifier determines where on the backbone to cut. See the figure below for where possible cuts are. Note that if no specifier is given, then the cute will occur at both N locations. That is, A:117-A:120 is equivalent to A:117N-A:120N.

QMDMD Cut

The `Exclude Side Chain` (list of strings) specifies side chain's to omit from the QM active site when using a series of residues. For example: 
"Residues" : ["A:117N-A:120", "A:204", "A:205N-A:206C"],
"Exclude Side Chain" : ["A:119"]

will exclude the side chain of residue A:119 from the active site, but include the backbone atoms of A:119. So a reverse cut is performed at the alpha-carbon, beta-carbon bond where the alpha-carbon is now capped with a hydrogen.

The Protonation sections is the exact same as the Protonation section in the dmdinput.json file. Though, it is necessary to specify the protonation states in both the DMD regions as well as the QM regions (so make sure to update both locations!)

As mentioned above, the Exclude Atoms (list of strings) allows you to specify chains, residues, or atoms to exclude from the QM calculation. These atoms are not frozen or replaced by anything else and are simply deleted from the coord file. For example

"Exclude Atoms": ["B", "C:400"]

will remove both the entirety of chain B as well as the residue C:400. This is particularly useful for multiple active sites where QM/DMD is only capable of performing QM calculations on a single active site.

The Substrate Chop allows the user to specify their own individual chopping options. While the name seems to indicate for substrate only, it can actually be used on an residues or molecule in the protein. Thus you can both specify chops within on the substrates, as well as a different chop on residues to remove unnecessary atoms in especicially large systems. An example is best illustrated on how this works.

"Substrate Chop": ["C:400:C105-:C:400:C107"]

This will chop between atoms C105 and C107 of residues 400 in chain C. It will then keep C105, freeze it, and cap it with a hydrogen in the direction of the C107 atom.

Additional Notes on the QM and DMD Sections

A few notes on the remaining parameters in the QM and DMD sections. First, the titratable feature should be turned off as the energies and passing protonation states between the DMD and QM regions is not well-defined right now. Furthermore, the Restrict Metal Ligands option should not be used during QM/DMD unless the user wishes to sample the different coordinations available. This is because during each QM/DMD iteration, the atoms near the metal will be re-evaluated, allowing for a dynamic coordination sphere, which is often not desired if the coordination environment is known a priori. Instead, users should specify the freeze ligating atoms in the Frozen atoms section and add dampening affects to the bonds to these frozen atoms in the Restrict Displacement section. This ensures that these parameters (which will appear in the inConstr file for DMD) will persist for the entirety of the QM/DMD simulation. Otherwise, the normal settings of Time and HEAT_X_C are fairly standard. Note that the Time represents the total DMD time spent during a single trajectory, not the total time it takes for the entirey of the DMD portion of the iteration.

For the QM section, cartesian coordinates should be used over internal redundant as this ensures TURBOMOLE does not rotate or translate the system to new coordinates. Additionally, do not worry about the freeze_atom parameter as this is taken care of during the QM chop. Also, the calculation will be set automatically for each step (ie, it will be set to ‘singlepoint’ and ‘geo’ automatically during the simulation). Lastly, the geo_iterations specifies the number of geometry optimization steps to take during a single QM/DMD iteration before continuing on to the next QM/DMD iteration. This is because it is not necessary to fully converge the system during each step, and it would be quite costly to do so.

Clone this wiki locally