Example of a molecular mechanics molecular 'topology' definition: the definition of an aspartate amino acid residue within the CHARMM22 protein MM potential function

 

·     Large molecules are typically divided into groups for the calculation of non-bonded interactions - each group bears an integral charge.

 

·     Note that the sidechain of the residue bears the whole of the negative charge of the residue.

 

·     Also note that the aliphatic hydrogen atoms are all assigned the same charge.

 

·     Note also the definition of which atoms are bonded to one another, and the specification of improper torsions which keep the peptide and carboxylate groups planar.

 

·     Each atom is listed by its name, atom type, and atomic charge. For example, NH1 indicates a peptide nitrogen, H a polar hydrogen, CT1 an aliphatic CH carbon, etc.

 

RESI ASP         -1.00    (-name and total charge)

GROUP

ATOM N    NH1    -0.47     !     |

ATOM HN   H       0.31     !  HN-N

ATOM CA   CT1     0.07     !     |   HB1   OD1

ATOM HA   HB      0.09     !     |   |    /

GROUP                      !  HA-CA--CB--CG

ATOM CB   CT2    -0.28     !     |   |    \

ATOM HB1  HA      0.09     !     |   HB2   OD2

ATOM HB2  HA      0.09     !   O=C

ATOM CG   CC      0.62     !     |

ATOM OD1  OC     -0.76

ATOM OD2  OC     -0.76

GROUP

ATOM C    C       0.51

ATOM O    O      -0.51

BOND CB CA  CG CB  OD2 CG

BOND N  HN  N  CA   C   CA  C +N

BOND CA HA  CB HB1  CB HB2

IMPR N   -C CA  HN  C CA +N O

IMPR OD1 CB OD2 CG

 

Potential functions for solid-state and ‘inorganic’ modelling

The MM potential functions described above work well for ‘organic’ molecules.

 

However, for modelling ionic solids, metals and semiconductors, it is usually not appropriate to use a localized covalent bond type of model.

 

Electrostatic interactions are very important, and 3-body and higher effects, and ionic polarizability, may need to be included (pairwise models of interaction energies don’t work very well).

 

Empirical potential functions have been developed which work well for solid-state systems (see e.g. Leach pgs. 236-245).

 

Next: geometry optimization