Date: Fri Jan 20 2017 - 09:34:05 CST
Dear NAMD Users
We would like to introduce a recently published article on calculation
of relative free energies.
The article includes a decoupling analysis in which the partition
functions of the transformed molecules are decomposed into two
partition functions. It enables to keep dihedral angle terms and
coupled bond angle terms at the separation point on, independently of
the function type, also when there is a bond junction at the
separation point between the shared and unshared submolecules. The
decoupling analysis is demonstrated in a relative solvation free
energy calculation of p-Cl and p-CH3 and the calculation is shown to
be more efficient since less terms are removed. In addition, it is
proved analytically that when capping the non-bonded potentials (soft
core technique) the integrated function (in TI) is monotonic (has
implications for robusteness). It is also shown mathematically that
when capping the non-bonded potentials and the two molecules are
simulated separately, the integrated function can be non-steep. It is
also explained how when the systems have rugged energy landscape they
can be equilibrated in the same sampling dimension.
Other important references there:
Ref. 34 (33), where it has been suggested to simulate the two
molecules separately and to cap the potential with accessible energies
of ~5kcal/mol (page 28, denoted by E_cutoff).
Ref. 35 where it has been suggested to remove terms only of the atoms
of the unshared submolecules (v1 abstract, v6 Fig. 7) and to cap
potential terms with accessible energies of 7kcal/mol (v1 Eq. (24)).
This reference has been split to the article on relative free energies
and the article on calculation of molecular free energies (Ref 38).
Ref. 38 which is about an exact calculation of molecular free energies
in a general environment. The calculations include the free energies
of bond stretching, bond angle, dihedral angle, bond junctions, and
complex structures. These free energies are also relevant for
restraints in binding.
Here is the link:
Thank you for your attention,
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:02 CST