Force Fields
Description
Our vision is that force fields will be readily available in a standardized format, well-studied and characterized, and well-understood for property or performance prediction.
Goals
Educate ourselves regarding other previous and on-going efforts that are similar to ours.
Obtain buy-in to strategic plan by key stakeholders.
Develop a test set of properties and molecules to bracket range of nonbonded interactions, i.e. increasingly polar, polarizable, hydrogen bonding donors and acceptors, ionic systems, increasing electrolyte concentration, pH, etc.
Members of the full team to provide list of desirable properties. Note: The FF team felt a small set of molecules with a long list of properties was preferable to the converse.
- Create the list of molecules.
- Identify existing citations for publications treating those molecules.
- Make the list of molecules and citations available on the web.
- Choose a test molecule on which to run the entire range of methods, from the quantum to classical to continuum to empirical, to serve as a benchmark and reference and to serve as a model to design an interface framework and proof-of-concept demonstration.
Choose the molecule. Done: ethylene oxide.
Apply each level of theory/simulation to calculate properties of the test molecule
Delineate all of the characteristics required to specify a force field (e.g., functional forms, combining rules, treatment of electrostatics, etc.). Identify/define a standard reference format for force fields to encourage formation of a publicly-accessible force field repository.
Complete (1) high-level design of interface to database and reference simulations, and (2) delineate the type of tools needed to enable users and developers to test and assess reliability and transferability of force fields and interpret results. Note: the team felt it was not practical to test all force fields developed into the future on all classess of molecules and properties, but that our role should be to provide the test framework, standards, and tools to enable users to test force fields and interpret results.
Complete high-level design of interface to database and reference simulations in a 1-page document to illustrate the concept
Delineate the type of tools needed to enable users and developers to test and assess reliability and transferability of force fields and interpret results
Create a database of force fields, populating for each one the characteristics identified.
Identify and publicize industrially-relevant classes of chemistries requiring improved descriptions and different formalisms and potential functions and make these assessments available via the web. Develop a sicklist and archive of known problems.
Lobby the community and journals to report the sensitivity of their results to parameters used when reporting new force fields. The goal is to include parameter sensitivity for every force field in the database.
Develop a series of test problems for developers to gauge transferability and how well force fields capture essential chemistry and physics to predict properties with the degree of reliability required by industry using generic established simulation methods. Need a systematic way for developers and users to quantify uncertainty. (Note: reliability replaced accuracy. Absolute statement of accuracy requirements is difficult because every case is different.)
Develop an online forcefield repository/tool whereby forcefield developers submit their force fields to the repository in a flexible, extensible, generic format; forcefields in the repository are then available and can be retrieved in a file format appropriate for several of the most popular simulation codes. Include a mechanism for user comments or ratings, as well as an indication of how many times each force field has been cited in the literature.
Other Resources
- Marcus Martin's list of force field references (http://towhee.sourceforge.net/references.html)
- CML (http://cml.sourceforge.net/)
- A. Pedretti, L. Villa, G. Vistoli, ATOM-TYPE DESCRIPTION LANGUAGE: A UNIVERSAL LANGUAGE TO RECOGNIZE ATOM TYPES IMPLEMENTED IN THE VEGA PROGRAM (http://www.ddl.unimi.it/papers/07/Teo.Chem.Acc.109_2003.pdf), Theor. Chem. Acc., Vol. 109 (4), 229-32 (2003).
- Junmei Wang, Wei Wang, Peter A. Kollman and David A. Case, "Automatic atom type and bond type perception in molecular mechanical calculations (http://dx.doi.org/10.1016/j.jmgm.2005.12.005)," Journal of Molecular Graphics and Modelling, Article in Press, Corrected Proof. (describes the Antechamber package)
- Antechamber
Team Notes
- On the topic of data to use for force field parameter adjustment, David Belashchenko (email received 25 Sept 05):"I think that it would be good to add the problem of databases of physical properties that could be used to calculate the parameters of inter-particle potentials. These are, for example, atomization energies, compressibility, diffraction data on the structure of liquids etc. Some of simulation methods use the diffraction data (structure factors, pair correlation functions) for the evaluation of effective inter-particle potentials and almost always there is lack of such data."
- On the topic of a force field database, Marcus Martin (email received 30 Sept 05): "I love the idea of a force field database, but you are going to be in for a lot of work if you attempt to start from scratch. I have probably implemented more force fields than anyone else and have learned a fair bit about a great number of force fields. The list of force fields already in Towhee can be found at (http://towhee.sourceforge.net/towhee_capabilities.html). All of these force fields are stored in Fortran subroutines as parameters and then output in my (ever changing) force field format by another subroutine. This same structure could be exploited if you wished to output into some format of your own choosing. I'd be happy to work with you folks as I would be very excited to be able to add force fields without having to implement them myself."
