Introduction to Molecular Dynamics: Theory and Applications in Biomolecular Modeling - Computational Modeling of Biological Systems - page 17

high sequence identity (>30%) between the reference and target protein is usually
required to ensure good quality of the homology model [ 85 ], and some expertise
may be needed in the refinement of the model [ 19 ].

5.1.2 pKa Calculation

Once a protein structure is chosen, the next step is to determine the protonation state
of each titratable protein residue. The protonation state of a residue, characterized by
its pKa value, is influenced by hydrogen bonding, desolvation effect and Coulombic
interactions in its local environment, and, therefore, can be very different from the
corresponding standard amino acid [ 32 ]. Many good programs are available for
predicting and assigning protonation states of protein residues, such as MCCE [ 3 , 4 ],
MEAD [ 10 ], PROPKA [ 9 ], and UHBD [ 82 ]. Performing pKa calculation using
at least one of these programs should become a routine in the preparation of a
simulation system.

5.1.3 Adding Water and Ions

Currently, the majority of MD simulations are performed using explicit water
molecules, although implicit solvent simulations have proven very useful in the
study of certain biomolecular systems [ 23 , 36 , 41 ]. In an explicit water simulation,
the number of water molecules needed is determined by the size of the simulation
box. As discussed in Sect. 3 , PBC are usually used to avoid the surface effect of
a finite-sized system. In these simulations, a rule of thumb is that the biomolecule
should never “see” its periodic image. This means that the simulation box has to be
large enough so that two neighboring periodic images of the molecule are separated
by at least the cutoff distance. In practice, a layer of water at least 10-15 ˚ Awide
is often added to each side of the protein. However, if the protein is expected to
undergo large conformational changes, such as unfolding, the simulation box should
be chosen large enough to accommodate the changes.
Apart from water, the buffer solution used in most biological experiments contain
various ions. Due to the limited availability of force field parameters, we cannot
hope to reproduce the exact experimental conditions in simulations. Nevertheless,
it is desirable to include ions, such as Na C ,Cl ,orK C in a system, to provide
a similar ionic strength in the simulation box as in the experiments. The added
ions should neutralize the net charge of the biomolecule, so that the total charge
of the simulation box is zero. However, the Ewald summation method described in
Sect. 3 introduces, by design, a homogeneous neutralizing background charge to the
system [ 56 ]. As a result, systems with nonzero net charges can be simulated without
any apparent error. Despite this result, it's generally considered a good practice to
keep the simulation system neutral, unless the goal is to simulate a charged system,
such as in the calculation of ionic solvation energy.