Software Information
Empirical Engines (Molecular Mechanics)
Several widely available programs perform molecular mechanics calculations. The major distinction between these programs are their force-field implementations. Some programs provide two or more force-fields.
The accuracy of modeled structures and properties depends heavily on the choice of energy calculation method. The first decision is whether to use molecular mechanics or a quantum chemical method. Given the decision to use molecular mechanics, a force-field that is suitable for the type of molecule being studied must be chosen.
Some of the more popular force-fields are listed below, together with the programs in which they are available and their chemical applicability:
- AMBER/Kollman (proteins, nucleic acids) - AMBER, Discover, SYBYL.
- CVFF (proteins, peptdies, organic molecules?) - Discover.
- CFF91 (proteins, peptides, organic molecules) - Discover.
- CHARMM (proteins, peptides) - CHARMM, CHARMm.
- Merck (organic molecules) - CHARMM, CHARMm.
- MM2/MM3 (organic molecules) - MM2, MM3.
- Scheraga (proteins, peptides) - ECEPP.
- Tripos (organic molecules) - SYBYL.
Notice that force-fields suitable for carbohydrates are noticeably absent.
CHARMm and Discover can be used via graphical interfaces or as stand-alone "batch" programs, although serious users will find numerous benefits to avoiding the graphical interfaces (i.e. QUANTA and Insight II) for many kinds of calculations.
When used on a stand-alone basis, these programs offer:
- Powerful scripting languages that exploit a greater range of capabilities than graphical interface programs support;
- Access from any terminal emulator program running on a personal computer (no need for a workstation).
The major disadvantage of using stand-alone molecular mechanics software is the additional expertise and setup time required.
Quantum Engines
Quantum engines are classified as either ab initio or semiempirical. ab initio engines are distinguished by their specific approaches to solving the Hartree-Fock equation, their available electron correlation methods (e.g. Configuration Interaction, Moller-Plesset, and variations thereof) and by their available basis sets.
Semiempirical engines are distinguished by the specific integrals that they neglect in solving the Hartree-Fock equation as well as their available basis and parameter sets.
Semiempirical HF Methods
Some of the major semiempirical methods are listed below, together with the programs in which they are available (note that some of these methods have historical value rather than modern day utility):
- Extended Huckel Theory (EHT) - FORTICON8.
- Complete Neglect of Differential Overlap (CNDO) and enhancements (CNDO/1, CNDO/2, CNDO/S, etc.).
- Intermediate Neglect of Differential Overlap (INDO).
- Modified INDO (MINDO) and enhancements (MINDO/2, MINDO/2', and MINDO/3) - AMPAC (MINDO/3 only).
- Michael Zerner's INDO (ZINDO) - ZINDO.
- Modified Neglect of Diatomic Overlap (MNDO) - AMPAC, MOPAC, Gaussian.
- Austin Model 1 (AM1) - AMPAC, MOPAC, Gaussian.
- Parametric Method 3 (PM3) - AMPAC, MOPAC, Gaussian.
- SemiChem Austin Model 1 (SAM1) - AMPAC. Explicitly treats d-orbitals.
QUANTA and SYBYL provide interfaces to MOPAC that automatically create the MOPAC input files and retrieve the output files for analysis. Insight II is similarly interfaced to both MOPAC and AMPAC.
Ab initio Methods
Some of the major ab initio programs are listed below:
- GAMESS
- Gaussian