PSU CHEM 408 - A Comparison of Conformational Energies

Unformatted text preview:

A Comparison of Conformational Energies Calculated by Several Molecular Mechanics Methods KLAUS GUNDERTOFTE H. Lundbeck A / S, Ottiliavej 9, DK-2500 Copenhagen-Valby, Denmark TOMMY LILJEFORS* and PER-OLA NORRBY Department of Medicinal Chemist y, Royal Danish School of Pharmacy, Universifetsparken 2, DK-2100 Copenhagen, Denmark INGRID PETTERSSON Department of Medicinal Chemistry, Astra Draco AB, PO Box 34, S-221 00 Lund, Sweden Received 7 Februa y 2995; accepted 3 August 1995 ,\BSTRACT Several commonly used molecular mechanics force fields have been tested for accuracy in conformational energy calculations. Differences in performance between the force fields are discussed for different classes of structures. MMFF93 and force fields based on the MM2 or MM3 functional form are found to perform significantly better than other force fields in the test, with average conformational energy errors around 0.5 kcal/mol. CFF91 also reaches this accuracy for the subset in which fully determined parameters are used, but it doubles the overall error due to use of estimated parameters. Harmonic force fields generally have average errors exceeding 1 kcal/mol. Factors influencing accuracy are identified and discussed. 0 1996 by John Wiley & Sons, Inc. different programs can be run on personal com- puters or small workstations. The problem today in routine work is not so much how to perform the calculation as what force field to use. A few years ago, two of us participated in a compafison of some of the most popular force fields at that time.' However, the last few years have seen the advent Introduction he molecular mechanics method has now be- T Come a s&mdard tool for Chemists. Sveral *Author to whom all correspondence should be addressed. Journal of Computational Chemistry, Vol. 17, No. 4, 429-449 (1996) 0 1996 by John Wiley & Sons, Inc. CCC 0192-8651 / 96 / 040429-21GUNDERTOFTE ET AL. of a plethora of new programs available on desk- top computers. We have therefore updated the previous study with some of the more popular current programs. All programs in this study have a graphics user interface for manipulating the molecules and per- forming the calculations. There are certainly large differences in what can be done with different packages. We have not made any attempt to grade the programs by user friendliness. This has been done elsewhere? We have solely evaluated the accuracy that can be obtained for calculation of relative conformational energies of organic molecules. To be able to reflect normal usage, we have adhered to default settings in the programs as long as no obvious errors were introduced by this. The test set of experimental energies is essen- tially the same as in the previous study.' Gas phase measurements of relative enthalpies have been used whenever possible. Computational Methods MACROMODEL Three different force fields in the program were tested: MM2*, MM3*, and AMBER*. Note that these differ in several respects from the parent force fields (MM2; MM3: and AMBER5). Some of the differences are discussed later. MacroModel version 4.0 (ref. 6) was used. Transition states were verified by normal mode analysis. Structures were converged to a gradient of 0.01 kJ mol-' A-l. CHEM3D Chem3D Plus version 3.1 (ref. 7) was used in this investigation. Minimizations were performed with the "MM2" force field. This implementation is based on, but not identical to, the MM2(87) force field? The main differences lie in an extended set of parameters and the ability to include centers with more than four bonds (e.g., metals). The de- fault convergence criterion was used (rms [root mean square] force 0.1 kcal mol-I A-'). CERIUS~ We used version 1.5.1 of Cerius' in this work: Three different force fields were used from this package. DREIDING 2.219 can be used for general organic and main group compounds, whereas UFF l.O1lo (the Universal Force Field) is able to perform calculations for the full periodic table. Also avail- able as a separate module is MMFF93 (Merck Molecular Force Field)? The default convergence criterion (rms force 0.1 kcal mol-' A-*) was used for all three force fields. The DREIDING force field uses charges from the parameter file to calculate the electrostatic con- tributions to the total energy. Most such charges in the force field are 0. However, halides have a charge of - 1. The use of these charges gives an overall negative charge of alkyl halides. To avoid this nonphysical situation, all charges were set to 0 in calculations with the DREIDING force field. UFF is recommended for use with a conforma- tionally dependent charge equilibrium method." The charge updating scheme does not check for consistency at the end of an energy minimization. Thus the energy would change if, after complete minimization, the charges were recalculated. Therefore, it was necessary to calculate charges iteratively and reminimize until the energy was constant within 0.01 kcal/mol. Cerius2 states that the UFF parameters were developed for use with the charge equilibrium scheme. However, in the original publication, the UFF authors say that the parametrization was done without charges."" Therefore, UFF was also tested with all charges set to zero. MMZ(9 1) The MacMimic/MM2 package'' was used. This implementation of MM2 is computationally identi- cal to the authentic I~M2(91).~ The default conver- gence criterion was used (AE < 0.00008 kcal/mol per atom). ALCHEMY Alchemy I11 for Macintosh (version 3.0) was used.13 This program utilizes a version of the Tripos force field.14 We used the default conver- gence criterion (rms gradient < 0.01 kcal/mol). Due to unreproducible results with charge calcula- tions: charges were not used in any calculation. This is consistent with the method used when the Tripos force field was validated by Clark et ~2.'~ 'We have been unable to find any published description of this particular force field. The version in Cerius' is identical to the 1993 version in CHAFWm. See ref. 57 for a more recent version of this force field. 'Building or importing ethanol in four different ways and minimizing when charges were active resulted in energies varying by 4 kcal/mol, despite all charges and atom types being identical. 430 VOL. 17, NO. 4COMPARI NG CON FORMATIONAL EN ERG1 ES PCMODEL The


View Full Document

PSU CHEM 408 - A Comparison of Conformational Energies

Documents in this Course
Load more
Download A Comparison of Conformational Energies
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view A Comparison of Conformational Energies and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view A Comparison of Conformational Energies 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?