28/10/2015

Computational Modelling in Physics,

Chemistry and Biochemistry

Molecular Modelling

Unit 3. Classic Model

Juan Torras, Univ. Politècnica Catalunya (2015)

Contents
 3.1. Molecular Interactions Modelling
 3.2 Bond Stretching
 3.3 Angle Bending
 3.4 Torsional Terms
 3.5 Improper Torsions
 3.6 Non-bonded interactions
 3.6.1 Electrostatic Interactions
 3.6.1.1 Multipole expansion
 3.6.1.2 Point-charge models
 3.6.1.3 Charge Derived from the Molecular Electrostatic Potential
 3.6.1.3.1 RESP charges
 3.6.1.4 Distributed Multipole Models
 3.6.1.5 Using Charge schemes to study inter-molecular interactions
 3.6.1.6 Polarization

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

1
 28/10/2015

Contents
 3.6.2 Van der Waals Interactions
 3.6.2.1 Dispersive Interactions
 3.6.2.2 Modelling Van der Waals Interactions
 3.6.3 Molecular Modelling of Hydrogen Bonding
 3.7 Force Field Models
 3.7.1 All Atom versus United Atom
 3.7.2 Force Field Parametrization
 3.7.3 Transferability
 3.8 Periodic Boundary Condition
 3.9 Truncation of non-bonded terms
 3.9.1 Van der Waals Potential
 3.9.2 Long-range electrostatic
 3.9.2.1 Ewald’s Summation
 3.10 Introduction to the GROMACS, AMBER, and CHARMM
Packages

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3. Classic Model
 Origin
 1930´s D.H. Andrews develops the mechanic model of the
molecule.
 A molecule is described as a set of masses connected by bonds
 If it is known the total system energy  Interactions between its
masses
 Any conformational change in the molecule  disturbance in the computable
energy
 Extended use of classic mechanics methodologies
 Computational limitations of quantum mechanics formalism to
study multiatomic systems
 The initial goal is the Minimization of the steric energy
 Description of van der Waals interactions and bond geometry
deformation

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

2
 28/10/2015

3. Classic Model
 System of N molecules
 Within Born-Oppenheimer approximation
 The rapid motion of electrons have been averaged out
 Hamiltonian of the system depends on coordinates
and momenta of each molecule

, ,…,
, ,…,
,

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3. Classic Model
 Kinetic energy
, ,

2
 Potential energy
 The energy is divided into terms depending on the coordinates of
individual atoms, pairs, triplets, etc
, , , …

 is the most important

  up to 10% of the lattice energy of argon
 Important in liquid phase
 Pairwise approximation  remarkably good description
 averaged three-body effects included within an “effective” pair potential

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3
 28/10/2015

3.1. Molecular Interactions Modelling

 Molecular mechanics
 Molecular interactions are calculated through classic potentials
 Inter and intramolecular potentials are approximately computed as the sum
of divers energy terms
 Inclusion of adjustable parameters
 It reproduce correctly the conformational behavior of organic molecules
 Model
 Atoms are represented by bulky spheres
 Mass and partial charge are defined
 Do not consider any specific electronic effect
 No electronic rearrangements can be studied
 chemical reactions, electronic transfer, etc.
 Conformational energy
 For each chemical species allegedly exist a set of natural bond angles and
distance to which the molecule tries to adapt its geometry, limited by its
structure and the present interatomic interactions

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.1. Molecular Interactions Modelling

 The Harmonic approach
 The classic potentials were originally developed for the vibrational
analysis of molecules
 Based on the classical formalism when a molecule moved away from an
equilibrium point, which is the potential energy minimum, the potential
energy of this new state can be expressed through a Taylor series:

1
Vpot V0
2
,

1
…
6
, ,

 V0 corresponds to the energy minimum, thus can be set to 0, as

reference point
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

4
 28/10/2015

3.1. Molecular Interactions Modelling

 The Harmonic approach
 If we assume that the potential only depends on the first non zero
term (third term) we obtain the so called harmonic approach

1 1
Vpot ≅
2 2
,
,
 At the equilibrium distance we assume a minimum of potential energy

 which generally remains valid when the displacement form the

equilibrium coordinates is not too big
 The proportional constant is named Force Constant
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.1. Molecular Interactions Modelling

 Correlating Force constant and experimental data
 Force constants would relate with experimentally measurable
magnitudes and they should reproduce the experimental values
 Stretching and bending  IR and Circular Dichroism
 Torsion  Microwaves Spectroscopy *
 Van der Waal radium  Apparent radii in crystal structures
 Data banks such as Cambridge Data Bank *
 Electrostatic charges  electron density measures
 X –Ray diffraction *
 If experimental data is not available then Quantum Chemistry
Ab Initio calculations need to be used to obtain reasonable to
good parameters.
 All parameters should be tested against a data set of known
molecules
* However most commonly obtained via fitting QM results

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

5
 28/10/2015

3.1. Molecular Interactions Modelling

 Force Field
 Set of classical analytical functions and empiric parameters that
allows computing the steric molecular energy of a determined
conformation.
 The empiric parameters are optimized from experimental and
theoretical properties
 Molecule geometry, conformational energy, formation heats, vibration
frequencies…
 The total energy of a molecular system is expressed as the
summation of different independent energy contributions

E E E

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.1. Molecular Interactions Modelling

 Force Field
 Energy of bonded Interactions
 Potentials associated with the chemical bonds (“the harmonic
expressions”)
E E E E
 Depending on the complexity of the harmonic expressions there are two
classed of Force Fields
 Those that use simple expressions with only one term are know as class I
 Common in biomacromolecules FF  AMBER, CHARMM, OPLS, GROMOS…
 and those that used complex summations (that include non harmonic terms
and crossed terms as well) are known as class II
 Common in organic chemistry FF  Allingher’s MMn series, UFF…
 Energy of non-bonded interactions
 Potentials that correspond to the rest of physical interactions within
molecular systems
E E E
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

6
 28/10/2015

3.1. Molecular Interactions Modelling

 Force Field
 Bonded interactions
 Energy potentials terms directly related
with the structure of the chemical bond Torsion

Bond
 Stretching stretching
 Term related with the extension and
contraction of the bond distance respect
to an equilibrium distance
Angle
bending
 Bending
 Flexion of the bond angle respect to the Non-bonded interactions
most stable value

 Torsion
 Related with the rotation around chemical
bonds (dihedral angle)

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.1. Molecular Interactions Modelling

 Force Field
 Non-bonded interactions
 Energy terms related with any intra-
and inter- molecular interaction that Torsion
does not implicate the chemical Bond
bond stretching
 Electrostatics
 Classic representation of the
interaction between the charge Angle
distribution of the atoms, which bending
includes polar and ionic interactions
Non-bonded interactions
 Van der Waals
 Energetic quantification of different
types of interactions, which share
their lack of electrostatic character
 The dispersion forces of London and
the repulsion between nuclei

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

7
 28/10/2015

3.1. Molecular Interactions Modelling

 Force Field

Michael Levitt, Nobel Lecture (2013)

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.2 Bond Stretching

 Morse potential
1
 It allows a bond dissociation ???
 Harmonic potential
 Quadratic equation (Hooke’s law) Hook
Morse
cubic
2
 Bond as a spring
 Minimum  equilibrium length
 Cubic potential 1 1.5
r0
2 2.5

 Allow a better approximation to Morse potential

1 …
2
 Anharmonic behavior  it takes more energy to compress a bond than it
does to stretch a bond the same distance. (Morse potential)
 Higher powers can be added to the harmonic equation to help simulate the
Morse function
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

8
 28/10/2015

3.3 Angle Bending

 Harmonic Hook function
 Is the preferred choice to compute angle bending energy
 Is possible higher powers to the first order for very strained systems
(i.e. cyclopropane)
Ebend
Atom
C sp3 63.21 110.63
C sp2 69.84 121.81
C sp1 64.44 180.00
C arom. 67.18 119.97
k1=25
k2=50 Gaussian Amber Force Field
C-C-C bending parameters
k3=75
k4=100

Angle
110 115 120 125 130 bending
θ

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.4 Torsional Terms

 Cosine function
 Allows description of energy barriers associated with the
rotation around chemical bonds
 The origin of this potential goes beyond the simple
interpretation in terms of the vdw repulsion between atoms
in eclipsed conformations
 The contribution of the vdw repulsion is only about10%
 Most of its origin has not classical explanation since the
potential energy barriers of rotation are mostly due to
quantum effects between bonding and non bonding orbitals
 Periodic function based on cosines of torsional angle with
two parameters: Amplitude , and Phase

1 cos

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

9
 28/10/2015

3.4 Torsional Terms

 Elaborated torsional barriers 1.5
V1=0.144; n=3; γ=0

 More complicated torsional barriers V2=1.175; n=2; γ=0

might be simulated by adjusting to a O‐C‐C‐O (V1+V2)
Fourier expansion of cosines, 1
reproducing satisfactorily QM rotation
Energy profiles.
 Most force fields use a single cosine with 0.5
appropriate barrier multiplicity, n 
Number of minimum points of energy
1 cos
0
0 60 120 180 240 300 360
θ
 Some use a Fourier expansion of cosines to correctly model the tendency
to adopt an specific conformation.
 O-C-C-O  adopt a gauche conformation (OH-CH2-CH2-OH)

1 cos ; cos
2

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.5 Improper Torsions

 Out-of-plane Bending motions
 Deformation of rigid bond geometries
 Movements out of the plane in molecules that usually present
conjugated π electrons
 Generally represented as a Fourier series or a sum of power
functions
1 cos 2 ;
2

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

10
 28/10/2015

3.6 Non-bonded interactions

 Two main interactions
 Electrostatic interactions
 Coulomb potential
 Multipole expansion

 Van der Waals interactions

http://proteopedia.org/wiki/index.php/Cation-pi_interactions

PNAS, cover photography, September 17, 2002; 99 (19)

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.1 Electrostatic Interactions

 Coulombic potential
 Is the most common way of representing electrostatic
term in Molecular Mechanics
 Interactions of pair of particles holding a partial charge

http://proteopedia.org/wiki/index.php/Cation-pi_interactions

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

11
 28/10/2015

3.6.1.1 Multipole expansion

 Some force fields compute the electrostatic term as
interaction between dipoles
 In this model the atom particles do not have net electric
charge
 The chemical bonds between different atom types present dipole
moments
 Magnitude depends on the kind of functional group each bonded
atoms belongs to.
 Electrostatic Energy is computed as the interaction energy
between charge-dipole, dipole-dipole, charge-quadrupole,
dipole-quadruple, quadrupole-quadrupole …

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.1.1 Multipole expansion

 The electrostatic potential at point P of a molecule made of two
charges at distance r from the origin
1
∅
4
 Expressing r1 and r2 in function of r and and expanding in a
Taylor’s series
1 cos Θ 3 cos 1
∅ …
4 2

P
 charge
 dipole r1 r2
Θ  quadrupole r
q1 θ q2
-z1 O +z2
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

12
 28/10/2015

3.6.1.1 Multipole expansion

 Electrostatic interaction between two molecules is calculated by
multiplying each charge by the potential tat that point and adding
the result for each charge
E ,
1 ′ 1
4
′
2cos cos sin sin cos …

 Efficient to calculate E between molecules

 Is not applicable at molecular distances
 Sphere of charges representing a molecule
ξ
should not intersect with other interacting
molecule spheres
 Not useful for intramolecular interactions
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.1.2 Point-charge models

 One charge minimum for each atomic nuclei
 Accurate representation may require charges placed out of
atomic nuclei  several water models

http://w3.iams.sinica.edu.tw/lab/jlli/thesis_andy/node8.html

 We assume that if sufficient point charges are used then

all of the electric moments can be reproduced and the
multipole interaction energy can be approximated
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

13
 28/10/2015

3.6.1.3 Charge Derived from the

Molecular Electrostatic Potential
 Atomic charges can be chosen to reproduce thermodynamic properties
 Only practical for small molecules
 Electrostatic properties are consequence of the distribution of the
electrons and the nuclei.
 Atomic charge is not an experimentally observable quantity
 Considerable debate about the best way to derive atomic charges
 Charges from Molecular Electrostatic Potential (MEP)
 MEP is an observable quantity  derived from wave function
 Obtain a set of partial charges that best describe the quantum mechanical
electrostatic potential at a series of points surrounding the molecule
 Symmetry condition can be imposed  charge of atoms symmetrically equivalents within the
molecule
 Trying to reproduce molecular dipole moment
 Minimize the function

 (QM potential) ; (Potential from set of points)

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.1.3.1 RESP charges

 Least squares algorithm derives atom centered charges
that best reproduces MEP Starting structure (PDB)

 Potential calculated on 4 shells of surfaces defined by {1.4,

1.6, 1.8, 2.0} x VDW radii (Singh & Kollman, J. Comput. Chem. 5
Perform Geometry
(1984)129 ) Optimization (HF-631G*)

 ESP at each point derived from QM

 Hyperbolic restraints on atoms  particularly buried Compute Electrostatic
Potential Charges (pop=mk)
atoms (RESP formalism: Bayly et al., J. Phys. Chem. 97 (1993) 10269 )
 Reduce high charges that can be assigned artificially in
standard electrostatic potential fitting methods Fit ESP charges (RESP)

MD simulation (AMBER)

Electrostatic potential
mapped in a isodensity
surface

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

14
 28/10/2015

3.6.1.4 Distributed Multipole Models

 Allows to represent the anisotropy of a molecular
charge distribution .
 Point charge, dipoles, quadrupoles, etc. Are distributed
throughout the molecule
 Distribute Multipole Analysis (DMA)
 Stone & Alderton, Mol. Phys. 56 (1985) 1047

 Calculates the multipoles from a quantum

mechanics wavefunction
 DMA depend strongly on molecular conformation
multipoles would have to be recalculated at each
conformational change
 Includes non-spherical, anisotropic effects
 i.a. lone pairs and pi electrons
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.1.5 Using Charge schemes to study

inter-molecular interactions
 Molecules containing systems
 Different orientations
 Edge-on, T-shaped and face-to-face
 Hunter & Saunders (1990)
 Benzene molecule with three charge on C
 +1 on C and two -1/2 chare above and
below
 Charge-charge interaction  Coulomb’s law
 Very simple method that
 - repulsion dominates  face-to-face
 - attraction dominates  edge-on
 - attraction dominates  offset stacked

Hunter & Saunders, JACS 112 (1990) 1024

Andrew R. Leach.
Molecular Modelling.
Prentice Hall. 2001

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

15
 28/10/2015

3.6.1.6 Polarization
 Usual FF’s do not account for electronic polarization of the
environment
 External electric field (neighboring molecules) induce a dipole in a
molecule,

-
- -- ++
- +
- +
 is the polarizability
 A third molecule can significantly reduce electrostatic interactions (size of
electric field) on the second molecule  Induction energy
 The energy of interaction between and E (induction energy) is
due to the work done in charging from 0  E
1
,
2
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.1.6 Polarization
 Main groups of methods including polarization effects in FF
 Fluctuating charge
 Based on the principle of electronegativity equalization: a charge flows between atoms
until electronegativities of the atoms become equalized. (CHARM)
 Drude oscillator (shell models)
 Atoms as a two-particle system: a charged core and a charged shell. The core and
shell are linked by a harmonic spring. (CHARM & GROMOS)
 Induced point dipole
 Classical point dipole approach  polarization energy is described as the
interaction between static point charges and the dipole moments they induce. (
OPLS/PFF, AMOEBA and AMBER ff02, ff02EP and ff02r1)
 Polarization treatment using continuum solvent (and solute)
 Continuum dielectric approach explicitly treat a molecule at the atomic level, and
the solvent or environment as a continuum medium, characterized by different
dielectric constants.
 Electronic polarization via quantum mechanical treatment or mixed QM/MM
 In the quantum mechanical region, electrons are explicitly treated and, so,
polarization effects are included in a natural way
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

16
 28/10/2015

3.6.2 Van der Waals Interactions

 van der Waals energy ( ) can be
expressed as the sum of two forces
 Short distance: repulsive force
 Long distance: attractive force between
atoms
 potential has a single
minimum
 The interatomic distance in which the
function is minimal correlates with the
Van der Waals radium
 Depth of the minimum well is related r0
with the polarizability of the atoms
(interaction strength
 The slope of the well is the atomic
hardness, i.e., the energy tolerance to
overlap atomic spheres

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.2.1 Dispersive Interactions

 Attractive contributions
 Due to dispersive forces (London forces)
 Instantaneous dipoles which arise during the electron density
fluctuations  can induce dipole in neighboring atoms

 Drude Model
 Molecules with two charges +q and –q, separated by a distance r
 Negative charge is vibrating in harmonic motion with respect to
stationary positive charge
 Predicts that the dispersion interaction varies as r -6
3
4 4
 Only considers dipole-dipole interaction
 Higher-order terms  dipole-quadrupole , quadrupole-quadrupole, …
…
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

17
 28/10/2015

3.6.2.2 Modelling Van der Waals

Interactions
 Lennard-Jones potential
 Most commonly used and easier to compute.
 Transferability  Lorentz-Berthelot mixing rules and
 However, it must be known that r -12 rises too rapidly compare to QM Energy
profiles (repulsive term)

EvdW 4ε ; ⁄2 ;

 Buckingham potential
 QM suggests exponential repulsion.
 This potential introduce an exponential repulsion term
 Hard to compute
 Adjusting parameter is a complex task
 Successfully computing crystals structures and ionic
species
 At very short distance  potential becomes strongly
attractive
https://en.wikipedia.org/wiki/Buckingham_potential

EvdW 6
6
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.6.3 Molecular Modelling of Hydrogen

Bonding
 Some force fields add an extra term
r
 Improves the accuracy in predicting the geometry of hydrogen-
bonding systems
 It requires that hydrogen bonds are identified before
performing the calculation
 other force fields just use a balance between electrostatic
and non-bonded terms

http://ww2.chemistry.gatech.edu/~lw26/structure/molecular_interactions/mol_int.html#E0

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

18
 28/10/2015

3.6.3 Molecular Modelling of Hydrogen

Bonding
 Water models
 Rigid models
 fixed atom positions (match known geometry
of water)
 only non-bonded interactions
 Flexible models:
 Include bond stretching and angle bending
 Reproduce vibration spectra
 Polarizable models:
 include explicit polarization term
 Enhances ability to reproduce water in different
phases and interaction between them
 TIP3P & TIP4P
 Jorgensen et al, J. Am. Chem. Soc.105 (1983) 1407
 TIP5P
 Mahoney & Jorgensen, J. Chem. Phys. 112 (2000) 8910
 TIP6P
 Chen & Smith, J. Chem. Phys. 126 (2007) 221101
 SPC/E
 Berendsen et al, J. Phys. Chem. 91, 6269 (1987)
http://www1.lsbu.ac.uk/water/water_models.html

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.7 Force Field Models

 The number of non-bonded interactions scales with
the square of the number of interactions sites.
 Interactions sites tends to be reduced
 Subsume H atoms into atoms to which they are bonded
 Two main system representation
 All Atom
 All atoms in the system are considered
 United Atom
 Methyl group modelled as a single “pseudo-atom”
 Coarse-Grained
 Chemical groups and units are modeled as a single “pseudo-atom”

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

19
 28/10/2015

3.7.1 All Atom versus United Atom

 United atom force fields
 Do not consider H atoms explicitly
 H atoms bonded to polar atoms (possible Hydrogen-bonds) are explicitly
represented
 Pros
 Several implementations: AMBER, CHARMM, GROMOS, …
 Reduce the effort by a factor ~2-3.
 Cons
 Chiral centers (in proteins) may be able to invert during a calculation
 Improper torsion term hast to be added to keep the side chain in the correct position
 The Van der Waals center is located at the carbon atom
 It does not reflect the presence of displaced hydrogen atoms
 The vdW term position should be offset slightly from C position

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.7.2 Force Field Parametrization

 PROTEINS:
 AMBER “Assisted Model Building with Energy Refinement” (Kollman/Case)
 http://ambermd.org/
 Main series: ff94, ff96, ff99, ff03, ff10
 Dihedral modifications: ff99sb, ff99sb-ildn, ff99sb-nmr, ff99-phi
 GAFF (Generalized AMBER force field) J. Comput. Chem. 25 (2004 ) 1157-1174
 OPLS “Optimized Potential for Liquid Simulation” (Jorgensen)
 OPLS-UA (united atom) J. Am. Chem. Soc., 110 (1988) 1657
 OPLS-AA (all atom) J. Am. Chem. Soc. 118 (1996) 11225-11236
 OPLS-AA/L (revised torsions)
 OPLS-2001, OPLS-/2005 (improved solvation free energies)
 CHARMM “Chemistry at Harvard Molecular Mechanics” (Karplus)
 http://www.charmm.org/
 CHARMM19 (united atom), CHARMM27 (all atom)
 CHARMM36 (carbohydrates)
 CMAP (two-dimensional dihedral corrections)
 CGenFF (General CHARMM force field) J. Comput. Chem. 31 (2010) 671–690
 AMOEBA “Atomic Multipole Optimized Energetics for Biomolecular
Applications”
 Contains polarizable point dipoles J. Phys. Chem. B 114 (2010) 2549–2564
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

20
 28/10/2015

3.7.2 Force Field Parametrization

 MARTINI force-field
 Four-to-one mapping
 Four type of particles
 Polar, Non-polar, Charged,
and Apolar.

Non-Bonded interactions

Marrink et al., J. Phys. Chem. B 111 (2007) 7812-7824

Bonded interactions

Ring particles (2 or 3 to 1 mapping)

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.7.2 Force Field Parametrization

 Force Fields are largely
empirical
 You get so many parameters:
Make them fit experiment!
 Ab initio calculations can be
used
 How does one get charges?
 Small molecules (representatives)
where to obtain parameter from
QM simulations
 FF should be properly
validated by testing against
experimental data
 i.e. Thermodynamic data (solvation
energy), equilibrium conformations
and vibrational frequencies Lindorff-Larsen et al. PLoS ONE 7(2) (2012) e32131

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

21
 28/10/2015

3.7.3 Transferability
 The modelling of systems containing mixtures of particles implies a bunch
of new parameters
 Those parameters have to be able to be “transferred” to new systems
 It is possible if all parameters are derived in the same way
 FF are developed to study:
 One atomic or molecular species under a wider range of conditions 
“general” FF
 An specific class of molecules: e.g. AMBER FF for proteins and nucleic acids
 “specific” FF
 The ability to transfer parameters form one molecule to another is
crucial
 The errors in the FF have to be balanced (similar error on different FF
terms)
 Transferability is helped by using the same parameters for as wide a
range of situations as possible.
 Bonding terms depend solely upon the atom types involved (C hybridization:
sp3,sp2,sp)
 vdW parameters could be independent of the atom type hybridization
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.8 Periodic Boundary Condition

 Computational capabilities are limited
 We only can simulate a small partition of
physical systems
 Surface effect
 Large fraction of molecules can be expected at
the surface rather than in the bulk
 Periodic Boundary Conditions (PBC)
 Eliminates surface effect-edge molecules
 Molecules have a mirror image
 Changes in one box are matched exactly in the
other boxes
 Particle motion in the central box
 Its periodic image in every one of the other boxes
moves with exactly the same orientation in exactly
the same way.
 Molecules exiting one wall re-enter through
the opposite!
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

22
 28/10/2015

3.8 Periodic Boundary Condition

 Within a system constituted by infinite
images of our simulation box,
 The non bonded energy would become an
infinite summation over an infinite array of
periodic images
 Minimum image convention (MIC)
∆x
 Avoid infinite summation of NBI
 Under PBC, inter-particle distances are
measured using MIC ∆y
 We must ensure rcutoff < L/2 so Lx
particles do not interact with multiple
images of neighbors
∆
∆ ∆

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.9 Truncation of non-bonded terms

 Non bonding potentials are computed as atom pairs
 The absolute number of computed pair energies increases exponentially
with the number of atoms present in the system
 Using PBC this problem becomes much more acute, even
when using MIC
 However, the effective force of the non bonding potentials depends on
the distance
 The interaction rapidly decrease as function of interatomic distance
 More acute in vdW interactions
 We can assume that beyond certain inter particle distance the energy contribution
could be negligible.

 We apply a cut–off distance (both real space or images)

beyond which non bonded pairs are not considered

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

23
 28/10/2015

3.9.1 Van der Waals Potential

 Cut-Off
 MCI conditions are defined by the cut off distance
 Any particle interacts only with those images that lie within the
boundary marked by the cut off distance
 Shift
 Lennard-Jones decays steadily over whole range and decays smoothly
to zero between user defined distances
 Smoothing function either at the energy or at the force expressions
 Steinbach et al. J. Comput. Chem. 15 (1994) 667-683

V r S r None shifted function

E
1
0
b
r
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.9.1 Van der Waals Potential

 Switch
 Lennard-Jones energy has a normal
behaviour out to a certain distance
 A correction function that forces energy
dropping to zero at cut off distance is
applied
 Steinbach et al. J. Comput. Chem. 15 (1994)
667-683
 Switching function applied to CHARMM
potentials

V r S r
1,
2 3
,

0,
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

24
 28/10/2015

3.9.2 Long-range electrostatic

 A cut-off introduces a
discontinuity in both the
Lennard-Jones
1.5
Coulomb

potential energy and the forces 1.0

E
1 qq
4 rij
near the cut-off value. 0.5

 If the potential energy function non 0.0

decays faster than 1/r3  makes -0.5

potential energy to diverge -1.0

1 2 3 4

 Coulombic interactions must be

treated specially 1
2 | |
 They characteristically extend they ,

effect through very long ranges n = n1 a1 + n2 a2 + n3 a3

 The decay of magnitude is very slow prime indicates omission of terms
with i = j when n = 0
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.9.2.1 Ewald’s Summation

 The Ewald sum splits up the slowly converging sum over the
Coulomb potential into the sum of two rapidly converging
series plus a constant term

 The real (direct) space sum ( ), the reciprocal (imaginary, or

Fourier) sum ( ), and the constant term ( ), known as the self-
term
Physical meaning
Each point charge is surrounded by a charge
distribution of equal magnitude and opposite
sign (Gaussian distribution)
A charge distribution of the same sign as the
original charge, and the same shape is also Toukmaji & Board, Comp. Phys. Comm. 95 (1996) 73-92
added
This cancelling distribution reduces the overall potential to that due to the original set
of charges (summed in reciprocal space)
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

25
 28/10/2015

3.9.2.1 Ewald’s Summation

 Self Term
 Correction term that cancels out the interaction of each of the introduced artificial
Gaussian counter charges with themselves
 corresponds to the sharp distribution of charge
 nmax, kmax, and κ control the convergence of the sums
 Usually, κ is chosen such that only the (n = 0) term is required to obtain the desired level
of accuracy, namely the minimum image convention within a spherical cutoff (Rc)

| |
1 | |
2 | |
,

| |
2 exp ⁄4
cos

, k = 2πm, and m sums over all integer translations of

the reciprocal lattice m = m1 a1* + m2 a2* + m3 a3*
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.10 Introduction to the GROMACS,

AMBER, and CHARMM Packages
 GROMACS
 GROningen MAchine for
Chemical Simulations
 http://www.gromacs.org/
 Primarily designed for
 Biochemical molecules
 i.e. proteins, lipids and
nucleic acids
 General systems
 i.e. polymers etc.
 Open source
 Serial, MPI and GPU binaries
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

26
 28/10/2015

3.10 Introduction to the GROMACS,

AMBER, and CHARMM Packages
 AMBER
 Assisted Model Building with Energy
Refinement
 http://ambermd.org/
 Open source (PMEMD under licence)

Solomon-Ferrer et al., WIRE: Comp. Mol. Sci., 3 (2013) 198-210

J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

3.10 Introduction to the GROMACS,

AMBER, and CHARMM Packages
 CHARMM
 Chemistry at HARvard Macromolecular Mechanics
 http://www.charmm.org/
 Primary focus
 Molecules of biological interest, i.e. peptides, proteins, prosthetic groups, small
molecule ligands, nucleic acids, lipids, and carbohydrates
 Systems in solution, crystals, and membrane environments
 Free for academic users
 Main Features
 Molecular Mechanics & Molecular  Reaction Path Determination
Dynamics
 Solvent and crystal simulations
 Genetic Algorithm
 Umbrella Sampling  Multi-body dynamics (MBOND)
 Miscellaneous Mean Field Potentials
(MMFP)  Quantum and Molecular
 Monte Carlo simulation package Mechanical Force Field (QM/MM)
 Free Energy Perturbation  NMR analysis facility
Calculations
 Normal Mode and Molecular  Reference Interaction Site Model
vibrational analysis facility (RISM)
J.Torras - UPC (2015) Molecular Modelling - Computational Modelling in Physics, Chemistry and Biochemistry

27
 28/10/2015

Computational Modelling in Physics,

Chemistry and Biochemistry

Molecular Modelling

Unit 3. Classic Model

Juan Torras, Univ. Politècnica Catalunya (2015)

28