The Basics: Getting Started With DivCon
Since quantum mechanics (QM) does not rely on atom types and is generally more physically accurate in comparison to molecular mechanics methods, getting started with QM is easier then one might expect. Below we list a number of key, “rules of thumb” for the application of QM to drug discovery. Many of these items are captured within the MOE/DivCon interface; however, if you prefer to run these methods outside of the MOE interface or if you are running DivCon as part of an X-ray refinement workflow, you will be required to keep these items in mind.
- QM is an all-atom method. This reality means that protons are required prior to simulation. Protonate3D within MOE is an excellent tool to add protons. Note: no automatic protonation tool will be 100% accurate 100% of the time. Therefore, once the protons have been added, you should double check the protonation states of key positions in the protein and especially the ligand.
- QM is generally more sensitive to the concept of “garbage-in/garbage-out.” What this means is if a structure has particularly bad or chemically incorrect contacts, bond lengths, etc., you may experience trouble. These problems can manifest themselves in longer calculation times, questionable energies, and poor predictions. Initial structure minimization using molecular mechanics methods is suggested, as is the use of docking software that optimizes or “refines” the poses generated.
- DivCon only requires simple input: the elements, the XYZ coordinates of those elements, and some sort of demarcation corresponding to the amino acids (in an enzymatic system for example). A powerful perception algorithm has been implemented in DivCon that will work for the vast majority of cases. In the unlikely event that it fails for some reason, you are provided with an opportunity to “tune” the charge using command line arguments.
- DivCon employs semiempirical QM that relies on Restricted Hartree-Fock methods which means that DivCon requires that the chemical system is closed shell. Generally for biological systems, this limitation is not an issue; however, when treating certain types of systems, it can create a problem. Generally, the perception algorithm will choose metal oxidation states that lead to closed shell systems. If the charge is incorrectly determined, you can change it using the command line; however, odd numbers of electrons can result at which point you will be warned. Unrestricted Hartree-Fock – which will allow for the treatment of open-shell systems – is being considered for a later release.
Basic Usage: MOE/DivCon, PHENIX/DivCon, and Beyond
The core QM engine of the Suite is the DivCon executable. Generally, the goal of QuantumBio is to integrate this software with other execution engines, graphical user interfaces, and so on. This way you are able to use these advanced software simulation tools in environments that you already find familiar. While this is a laudable goal and it is one that allows the user to perform most of the functions of the DivCon suite, there are some limitations. Specifically, the user is only able to run the calculations, workflows, and protocols that have been implemented by QuantumBio staff. The DivCon command line executable is also provided for those cases where the advanced user may want to run very specific types of calculations. The following contextual list is provided in order to help navigate the Manual for your particular needs.
MOE/DivCon: The MOE/DivCon interface supports X-ray Refinemnt (with additional functions on the way). The popular MOE platform allows the user to correctly and completely prepare a PDB structure as per the above noted “rules of thumb.” The user can read in the PDB (or Mol2 or Maestro or whatever) file using MOE’s supported parsers; protonate the structure using Protonate3D; optimize the structure using OPLS-AA, MMFF, AMBER, and so on; and finally run a number of tests on the structure to make sure that it is ready for QM treatment. Taken together, this is a pretty complete package.
PHENIX/DivCon: The DivCon executable has been integrated with the Python-based Hierarchical ENvironment for Integrated Xtallography (e.g. PHENIX) for X-ray refinement package. PHENIX is one of the most popular refinement platforms on the market, and this integration brings the power and versatility to the package. At its core, the refinement method championed by PHENIX uses a mixture of stereochemical restraints and experimental data to arrive at an acceptable X-ray model. With PHENIX/DivCon, some or all of the stereochemical restraints are replaced with gradients from the chosen semiempirical Hamiltonian (e.g. AM1, PM3, or PM6) at each refinement cycle. This QM data is generated in real time as the refinement progresses, and can address particularly exotic systems such as metal containing systems, bound ligands, strange ligand chemistry, and so on.
qmechanic (DivCon Command Line Interface): Similar to a classic command line-based package, qmechanic can use arguments provided on the command line to for structure optimization, energy calculation, nuclear magnetic resonance, and so on.