Recently, July 2013, a Molecular Mechanics correction was added to RM1, PM6, and PM7.  MM corrections have been available for MNDO, AM1, and PM3 for a long time.  Because of the recent change to RM1, PM6, and PM7, keyword MMOK must be used if the molecular mechanics correction is wanted.  MMOK is only needed if peptide bonds are present. Eventually MMOK will become the default, and the keyword will become redundant.

If the molecular mechanics correction is not wanted, then NOMM should be used.

If the system contains a peptide linkage, then when a molecular mechanics correction is made the barrier to rotation about the peptide bond is increased (to 14.00 kcal/mol in N-methyl acetamide).

The peptide linkage consists of the group -CO-NH-.  In most instances the O=C-N-H torsion or dihedral angle is near to 180° and in a few instances the angle is roughly 0°, in other words, most peptide bonds are more-or-less flat.  In proteins, because of environmental effects, the deviation of the angle from these ideal angles can be greater.   The most important deviations from planarity occur when a peptide bond changes from cis to trans or vice versa.  During such reactions the torsion angle goes through 90°.  In the case of N-methyl-acetamide, the barrier to rotation has been calculated, using highly accurate ab-initio methods, to be about 14 kcal/mol.  Semiempirical methods predict a smaller barrier.  To correct this fault, a small molecular mechanics correction is added to the calculated heat of formation; this correction has the form:

ΔHf' = ΔHf +c.sin(θ)2

where c is a constant depending on the semiempirical method, and f is the O=C-N-H torsion angle.  Obviously, if the torsion angle is 180° or 0°, i.e., the default peptide bond angle, then no correction will be made, but for all other angles there is an energy penalty that biases the peptide bond towards planarity.

In practice, this MM correction to the peptide bond produces an insignificant bias towards planarity in proteins, because most peptide bonds in proteins are almost planar anyhow, and a significant change in barrier heights when reactions involving twisting a peptide bond are modeled.