Tooling Up for a Molecular Modeling Project
Due to the seeming user-friendliness of many molecular modeling programs and the easy access that many people have to powerful hardware, it is tempting to just sit down at the computer and start working. As for laboratory experiments, molecular modeling "experiments" need to be planned out from start to finish. Considerations include:
- Goals:
- Visualization only?
- Scratchpad for ideas?
- Publication quality results?
- Model definition:
- Composition:
- Single molecule.
- Multi-subunit molecule.
- Molecular complex (e.g. DNA-protein).
- Representative fragment (e.g. binding site).
- State:
- Starting structure and its source:
- De novo (usually small organics or oligos).
- X-ray/neutron crystal structure.
- NMR NOE constraints.
- Ionizable groups (i.e. pH and pKa considerations).
- Electronic (i.e. ground or excited).
- Environment:
- Water.
- Lipid membrane.
- Binding pocket.
- Counter-ions.
- Strategies:
- Modeling and analysis method(s).
- Target properties.
- Useful data representations.
- Tools:
- Software and hardware.
The choice of software can be based on the following criteria:
- Practicality (is the software applicable to the problem at hand?)
- Generation, storage, and analysis of all essential data (breadth and depth of available tools).
- Reasonable computational time and resource requirements.
- Scientific quality (are the results believable?)
- Accuracy of computed data.
- Ability to validate by experiment or by agreement between different computational methods.
- Convenience (how many hoops to jump through?)
- Availability of interfaces to other software.
- User friendliness.
- Software command scripts.
- Input requirements:
- Parameters (usually supplied, but may be incomplete or unsatisfactory).
- Molecular structure information.
- Software-specific data files.
- Expermental data.
General Software Information
The Anatomy of a Molecular Modeling Software Package
There are three fundamental operational areas addressed by molecular modeling software: computation, geometry, and graphics. Each of these areas is a subject unto itself. While briefly summarized below, they are considered in greater depth elsewhere in this Guide. The scope of a molecular modeling program is determined by the nature of its capabilities in each of these fundamental areas. Fundamental capabilities can be hooked together in a single program to perform complex operations such as molecular dynamics or ligand-binding site docking.
Commonly used combinations of capabilities include:
COMPUTATION + GEOMETRY + GRAPHICS (e.g. SYBYL)
COMPUTATION + GEOMETRY (e.g. Discover)
GEOMETRY + GRAPHICS (e.g. NAMOT)
GEOMETRY only (e.g. DGEOM)
GRAPHICS only (e.g. RasMol)
A molecular modeling program consists of a user interface and one or more "engines" that calculate (i.e. predict or simulate) molecular structures and properties.
Molecular Modeling Software Interfaces
The simplest kind of interface uses arguments of the run command to pass user-supplied instructions to the program, performing one operation each time the program is run (e.g. "program -i 1crn.pdb -o output.dat"). "Babel", a program that translates between different types of molecular structure file formats, is one example of this type of interface.
More complex interfaces are based on interactive commands or command languages (issued while the program is running) to control a program's functions.
Many programs allow a sequence of commands (command stream) to be stored in a file (called a "script") that can be used to automatically operate the program. MolScript, a molecular graphics program, can only be operated from a command script, while "CHARMM", a molecular mechanics/dynamics program, can be operated using interactive commands or from a command script.
The third type of interface is based on an interactive graphical display (e.g. a set of icons or widgets and pull-down menus) that allows a program to be controlled using a pointing device (e.g. mouse). Graphical displays typically provide capabilities for interactively viewing and manipulating molecular structures. QUANTA, a multi-purpose molecular modeling program, exemplifies this kind of interface.
Dual command/graphical interfaces are common in molecular modeling programs, wherein command scripts can be used to perform complex or repetitive operations and a graphical interface can be used for freestyle interaction with the program.
Molecular Modeling Engines
The actual structure/property calculations are performed by the engines of a molecular modeling program. Many kinds of molecular structure and property calculations require two or more closely coupled engines. For example, energy minimization can be performed using a combination of molecular mechanics (computation) and minimization (geometry) engines.
Specialized (i.e. one engine) programs can sometimes be used in tandem to calculate molecular structures and properties or analyze the resulting data. For example, structures calculated using CHARMM can be displayed using a molecular graphics program like RasMol. In these cases, the user is often left to deal with data compatibility problems between the programs.
The utility, "Babel", that interconverts many kinds of molecular modeling data file formats, can be of great help in this regard.
Multi-purpose programs offer the convenience of integrated capabilities operating under a uniform interface, but the capabilities of specialized stand-alone programs are sometimes more extensive and/or better.
Computation Engines
Computation engines are used to calculate molecular energies (e.g. heat of formation, electronic, electrostatic, total, etc.) and properties associated with energy (e.g. charge distribution, dipole moment, electrostatic potential, etc.). The energy calculation capabilities vary between the three different types of engines (i.e. empirical, semiempirical, and ab initio) and their specific implementations. See the NIH Molecular Modeling Home Page software list for information concerning specific programs containing computation engines.
Geometry Engines
Geometry engines are used to build, manipulate, and measure three-dimensional molecular structures. Structures can be built from standard bond angle and length data or using distance geometry-based methods. When coupled to a computational engine, geometry engines (e.g. minimization, sampling, dynamics, Monte Carlo, and perturbation) can be used to evaluate structural variation as a function of energy and to perform statistical mechanics analyses. Different combinations of computational and geometry engines can be used to study such problems as conformational energy surfaces and transitions, reaction pathways, and intermolecular interactions (i.e. complexes). Other geometry engines can be used to fit molecular structures for the purpose of comparison. See the NIH Molecular Modeling Home Page software list for information concerning specific programs containing geometry engines.
Graphics Engines
Graphics engines provide a means of visualizing molecular structure and property data. When coupled to other engines, molecular graphics can be used for setting up calculations and for visual interpretation of computational results (e.g. surface mapping and contouring, molecular dynamics trajectories). Some molecular graphics programs can be used to create visually realistic (rendered) pictures of molecular structures and representations of properties. Some programs can generate color PostScript instructions or other forms of high resolution images for publications or slides. See the NIH Molecular Modeling Home Page software list for information concerning specific programs containing graphics engines.
