scpism (c38b2)
Screened Coulomb Potentials Implicit Solvent Model (SCPISM)
The SCPISM is a continuum model of solvation that treats implicitly the
effects of water. The model is based on screened Coulomb potentials (SCP).
The screening function are derived from the Lorentz-Debye-Sack theory of polar
liquids. In the present implementation the model incorporates a continuum
description of electrostatics and calibration of hydrogen bond energies.
The model uses (optionally) a cavity term to account for non-polar solvation energy.
The current implementation is suitable for dynamics (plain or Langevin) simulations,
energy evaluation and minimization of peptides and proteins. The models is to be
used in combination with the all-atom representation (either PAR22 or CMAP)
In the current implementation there is one parameter per atom type.
The original parameterization was done based on experimental solvation energies
of amino acid side chain analogs [2] (the parameterization was not based on reproducing
PB results). Hydrogen bonding strength is treated independently [4]. This was shown to
be important for ab initio structure prediction and stabilization [3].
* Menu
* Syntax | Syntax of the SCPISM commands
* Background | An introduction to the SCPISM (see also URL)
* References | Useful references (see URL)
* Example | Input file
Top
SCPISM commands
An effort was made to minimize the number of input options available
to the user (i.e., no parameters are allowed to be modified from an input
file since the physics of the system was already incorporated into the model
and, then, hardwired into the algorithm). To activate the model the following
command line is used:
SCPIsm [UISM int] (1)
where UISM is the unit number for reading the SCP parameters. These
parameters are stored in scpism.inp and must be opened for reading
in the usual way before the model is requested (see example below), i.e.,
OPEN READ UNIT int CARD NAME "scpism.inp"
Once the model is requested, any options for energy, minimization
and dynamics calculations are supported. Electrostatic
interactions are truncated using a shift function (any other
specification of cutoff of electrostatics is automatically disabled
once the SCPISM is requested; a shift function was shown to be the best
option in the SCPISM.
Any restraining option that is introduced as an additional term in the
potential is supported. Use of the CONS FIX constraints is discouraged
because they are inconsistent with the way the self-energies are
calculated: the calculation of atom effective radii in the SCPISM requires the
positions of all the atoms around each atom to be available, in particular
the positions of the atoms to be fixed.
It is possible (but not required) to deactivate the model once any
of the required tasks (energy evaluation, minimization or dynamics) is
completed. To exit the model use
SCPIsm END
in this case CHARMM will return to the default (vacuum) electrostatics
options (the model can be turned on again by using the command line (1) above).
Note that, although the current implementation of the SCPISM describes
only electrostatic and HB effects, to be consistent with earlier applications,
a cavity term proportional to the total solvent accessible surface area (SASA)
can be requested by using the subcommand HYDRophobic in the
command line (1) above (see example). If this term is not activated,
other model accounting for hydrophobic interactions should be used.
The CPU time of the serial version of the SCPISM has been improved to be only
1.5 times slower than vacuum (default compilation optimization). Since
version c34 the models has been parallelized (MPI routines added to
the scpism.src source code). But from c36 onward the performance has been improved
substantially, and is currently as follows (Milan Hodoscek, 2011))
Benchmark for ATPase (>15000 atoms, 100 steps dynamics) using 32 core AMD box:
CPUs scpism (charmm c34 and 35) | scpism (charmm 36)
time (sec) speedup efficiency | time (sec) speedup eff.
1 61.7 1.00 100% | 61.8 1.00 100%
2 43.5 1.42 71% | 33.0 1.87 94%
4 58.4 1.06 26% | 17.0 3.63 91%
8 105.4 0.59 7% | 9.0 6.87 86%
16 218.4 0.28 2% | 5.1 12.12 76%
32 453.4 0.14 0% | 3.3 18.73 59%
SCPISM commands
An effort was made to minimize the number of input options available
to the user (i.e., no parameters are allowed to be modified from an input
file since the physics of the system was already incorporated into the model
and, then, hardwired into the algorithm). To activate the model the following
command line is used:
SCPIsm [UISM int] (1)
where UISM is the unit number for reading the SCP parameters. These
parameters are stored in scpism.inp and must be opened for reading
in the usual way before the model is requested (see example below), i.e.,
OPEN READ UNIT int CARD NAME "scpism.inp"
Once the model is requested, any options for energy, minimization
and dynamics calculations are supported. Electrostatic
interactions are truncated using a shift function (any other
specification of cutoff of electrostatics is automatically disabled
once the SCPISM is requested; a shift function was shown to be the best
option in the SCPISM.
Any restraining option that is introduced as an additional term in the
potential is supported. Use of the CONS FIX constraints is discouraged
because they are inconsistent with the way the self-energies are
calculated: the calculation of atom effective radii in the SCPISM requires the
positions of all the atoms around each atom to be available, in particular
the positions of the atoms to be fixed.
It is possible (but not required) to deactivate the model once any
of the required tasks (energy evaluation, minimization or dynamics) is
completed. To exit the model use
SCPIsm END
in this case CHARMM will return to the default (vacuum) electrostatics
options (the model can be turned on again by using the command line (1) above).
Note that, although the current implementation of the SCPISM describes
only electrostatic and HB effects, to be consistent with earlier applications,
a cavity term proportional to the total solvent accessible surface area (SASA)
can be requested by using the subcommand HYDRophobic in the
command line (1) above (see example). If this term is not activated,
other model accounting for hydrophobic interactions should be used.
The CPU time of the serial version of the SCPISM has been improved to be only
1.5 times slower than vacuum (default compilation optimization). Since
version c34 the models has been parallelized (MPI routines added to
the scpism.src source code). But from c36 onward the performance has been improved
substantially, and is currently as follows (Milan Hodoscek, 2011))
Benchmark for ATPase (>15000 atoms, 100 steps dynamics) using 32 core AMD box:
CPUs scpism (charmm c34 and 35) | scpism (charmm 36)
time (sec) speedup efficiency | time (sec) speedup eff.
1 61.7 1.00 100% | 61.8 1.00 100%
2 43.5 1.42 71% | 33.0 1.87 94%
4 58.4 1.06 26% | 17.0 3.63 91%
8 105.4 0.59 7% | 9.0 6.87 86%
16 218.4 0.28 2% | 5.1 12.12 76%
32 453.4 0.14 0% | 3.3 18.73 59%
Top
Structure of the parameter input file:
All the parameters required for the model are atom-type based and are
collected in a single file. Determination of these parameters was described in
[2,3]. The parameters have been optimized in the context of the all-atom force
field.
SCPISM parameter file has the following format (note that not all fields
in this file are parameters that controls the electrostatics):
H 0.4906 0.6259 0.5000 0.7004 0.3700 0.0052 PH ! polar H
HC 0.5300 9.6746 0.5000 0.7280 0.3700 0.0052 PH ! N-ter H
HA 0.5300 0.5000 0.5000 0.7280 0.3700 0.0052 ! nonpolar H
HT 0.4906 2.3800 0.5000 0.7004 0.3700 0.0052 PH ! TIPS3P WATER HYDROGEN
HP 0.5274 0.5000 0.5000 0.7262 0.3700 0.0052 ! aromatic H
HB 0.4858 0.5000 0.5000 0.6970 0.3700 0.0052 ! backbone H
HR1 0.4651 0.5000 0.5000 0.6820 0.3700 0.0052 ! his he1, (+)his HG,HD2
HR2 0.4580 0.5000 0.5000 0.6768 0.3700 0.0052 ! (+) his HE1
Col. 1: Atom type defined in PAR22
Col. 2: Alpha_i controls slope of D(r) around atom-type i
Col. 3: for PH interaction with PA this value controls the effective radius
of PH to modulate hydrogen bonding strength (see URL); the
increase or decrease of the PH effective radius is defined by the
product Col.3(PH)*Col.3(PA) As a rule, the larger the value of
this product, the weaker the HB interaction.
Col. 4: Extension of effective radius R_iw to obtain R_ip, i.e.,
R_ip = R_iw + Col.4 (see [1]); only one value for atoms considered
Col. 5: SQRT(alpha_i); this is needed for alpha_ij=alpha_i alpha_j (see [1,2])
Col. 6: covalent radius for each atom (only 5 values considered, i.e.,
for C,N,O,S,H)
Col. 7: gamma_i in hydrophobic energy = sum over i of gamma_i SASA_i (note
that all coefficients are equal in this first release)
Col. 8: denotes atoms involved in H-bonding (PH = polar proton; PA =
proton acceptor)
Theoretical background
The SCPISM is based on a superposition of screened potentials. It
uses screening functions D(r) that modulate the potential phi(r) rather
than on dielectric functions eps(r) that modulate the electric field E(r).
Both D(r) and eps(r) are sigmoidal functions of r. Once eps(r) is known
from theory or experiments (see [2,5,6]) D(r) is obtained by integration.
Based on these results, the SCPISM uses atom type-dependent sigmoidal
functions in the context of the all-atom representation.
In the SCPISM the standard electrostatic component of the force field
is replaced by terms that describe both the electrostatic interaction energy,
and the self energy. The screening functions are continuous functions of the
position and describe a dielectric medium that permeates all of space. For the
solvated protein, D(r) approaches bulk screening only far from the protein
(see [2,5,6] for discussion). Therefore, the SCPISM does not introduce either
an internal or an external dielectric constant and, then, there is no boundary
that separates the protein from the solvent. The form of the screening is
derived from basic theory of polar solvation. The model is being constructed
incrementally to incorporate all the relevant physical of solvation that are
removed when the explicit solvent is eliminated. Most important among these
are hydrogen-bond interactions at interfaces, water-exclusion effects, and
other solvent-induced forces [6-8].
Because the effective screening functions that characterize the overall
modulation of the electrostatics are obtained from properties of bulk solvent,
short-range interactions characterizing hydrogen bonding (HB) must be corrected
[3,4]. To obtain a reasonable representation of HB strength in solvent medium,
all individual HB interactions among amino acids pairs are individually calibrated
via the self-energy terms (by adjusting the affective atom radii of polar
hydrogens [3,4]). The stabilization of HB energies is carried out based on charge
states of the interacting groups, as well as on hybridization states of donor
and acceptor atoms.
Structure of the parameter input file:
All the parameters required for the model are atom-type based and are
collected in a single file. Determination of these parameters was described in
[2,3]. The parameters have been optimized in the context of the all-atom force
field.
SCPISM parameter file has the following format (note that not all fields
in this file are parameters that controls the electrostatics):
H 0.4906 0.6259 0.5000 0.7004 0.3700 0.0052 PH ! polar H
HC 0.5300 9.6746 0.5000 0.7280 0.3700 0.0052 PH ! N-ter H
HA 0.5300 0.5000 0.5000 0.7280 0.3700 0.0052 ! nonpolar H
HT 0.4906 2.3800 0.5000 0.7004 0.3700 0.0052 PH ! TIPS3P WATER HYDROGEN
HP 0.5274 0.5000 0.5000 0.7262 0.3700 0.0052 ! aromatic H
HB 0.4858 0.5000 0.5000 0.6970 0.3700 0.0052 ! backbone H
HR1 0.4651 0.5000 0.5000 0.6820 0.3700 0.0052 ! his he1, (+)his HG,HD2
HR2 0.4580 0.5000 0.5000 0.6768 0.3700 0.0052 ! (+) his HE1
Col. 1: Atom type defined in PAR22
Col. 2: Alpha_i controls slope of D(r) around atom-type i
Col. 3: for PH interaction with PA this value controls the effective radius
of PH to modulate hydrogen bonding strength (see URL); the
increase or decrease of the PH effective radius is defined by the
product Col.3(PH)*Col.3(PA) As a rule, the larger the value of
this product, the weaker the HB interaction.
Col. 4: Extension of effective radius R_iw to obtain R_ip, i.e.,
R_ip = R_iw + Col.4 (see [1]); only one value for atoms considered
Col. 5: SQRT(alpha_i); this is needed for alpha_ij=alpha_i alpha_j (see [1,2])
Col. 6: covalent radius for each atom (only 5 values considered, i.e.,
for C,N,O,S,H)
Col. 7: gamma_i in hydrophobic energy = sum over i of gamma_i SASA_i (note
that all coefficients are equal in this first release)
Col. 8: denotes atoms involved in H-bonding (PH = polar proton; PA =
proton acceptor)
Theoretical background
The SCPISM is based on a superposition of screened potentials. It
uses screening functions D(r) that modulate the potential phi(r) rather
than on dielectric functions eps(r) that modulate the electric field E(r).
Both D(r) and eps(r) are sigmoidal functions of r. Once eps(r) is known
from theory or experiments (see [2,5,6]) D(r) is obtained by integration.
Based on these results, the SCPISM uses atom type-dependent sigmoidal
functions in the context of the all-atom representation.
In the SCPISM the standard electrostatic component of the force field
is replaced by terms that describe both the electrostatic interaction energy,
and the self energy. The screening functions are continuous functions of the
position and describe a dielectric medium that permeates all of space. For the
solvated protein, D(r) approaches bulk screening only far from the protein
(see [2,5,6] for discussion). Therefore, the SCPISM does not introduce either
an internal or an external dielectric constant and, then, there is no boundary
that separates the protein from the solvent. The form of the screening is
derived from basic theory of polar solvation. The model is being constructed
incrementally to incorporate all the relevant physical of solvation that are
removed when the explicit solvent is eliminated. Most important among these
are hydrogen-bond interactions at interfaces, water-exclusion effects, and
other solvent-induced forces [6-8].
Because the effective screening functions that characterize the overall
modulation of the electrostatics are obtained from properties of bulk solvent,
short-range interactions characterizing hydrogen bonding (HB) must be corrected
[3,4]. To obtain a reasonable representation of HB strength in solvent medium,
all individual HB interactions among amino acids pairs are individually calibrated
via the self-energy terms (by adjusting the affective atom radii of polar
hydrogens [3,4]). The stabilization of HB energies is carried out based on charge
states of the interacting groups, as well as on hybridization states of donor
and acceptor atoms.
Top
References
[1] S A Hassan, E L Mehler, D Zhang and H Weinstein, Molecular Dynamics
Simulations of Peptides and Proteins with a Continuum Electrostatic Model
based on Screened Coulomb Potentials; Proteins 51, 109 (2003)
[2] S A Hassan, F Guarnieri and E L Mehler, A General Treatment of Solvent
Effects based on Screened Coulomb Potentials; J Phys Chem B 104, 6478
(2000)
[3] S A Hassan, F Guarnieri and E L Mehler, Characterization of Hydrogen
Bonding in a Continuum Solvent Model; J Phys Chem B 104, 6490 (2000)
[4] S A Hassan, Intermolecular Potentials of Mean Force of Amino Acid Side Chain
Interactions in Aqueous Medium; J Phys Chem B 50, 19501 (2004)
[5] S A Hassan and E L Mehler, From Quantum Chemistry and the Classical
Theory of Polar Liquids to Continuum Approximations in Molecular Mechanics
Calculations; Int. J. Quant. Chem. 102, 986 (2005)
[6] S A Hassan, Liquid Structure Forces and Electrostatic Modulation of
Biomolecular Interactions in Solution; J. Phys. Chem. B 111, 227 (2007)
[7] S A Hassan and E L Mehler, A Critical Analysis of Continuum Electrostatics:
The screened Coulomb potential-implicit solvent model and the study of
the alanine dipeptide and discrimination of misfolded structures of
proteins; Proteins 47, 45 (2002)
[8] S A Hassan and E L Mehler, Modeling Aqueous Solvent Effect through Local
Properties of Water, in Modeling Solvent Environments: Applications to
Simulations of Biomolecules, Ed. M Feig, Wiley-VCH (2010)
References
[1] S A Hassan, E L Mehler, D Zhang and H Weinstein, Molecular Dynamics
Simulations of Peptides and Proteins with a Continuum Electrostatic Model
based on Screened Coulomb Potentials; Proteins 51, 109 (2003)
[2] S A Hassan, F Guarnieri and E L Mehler, A General Treatment of Solvent
Effects based on Screened Coulomb Potentials; J Phys Chem B 104, 6478
(2000)
[3] S A Hassan, F Guarnieri and E L Mehler, Characterization of Hydrogen
Bonding in a Continuum Solvent Model; J Phys Chem B 104, 6490 (2000)
[4] S A Hassan, Intermolecular Potentials of Mean Force of Amino Acid Side Chain
Interactions in Aqueous Medium; J Phys Chem B 50, 19501 (2004)
[5] S A Hassan and E L Mehler, From Quantum Chemistry and the Classical
Theory of Polar Liquids to Continuum Approximations in Molecular Mechanics
Calculations; Int. J. Quant. Chem. 102, 986 (2005)
[6] S A Hassan, Liquid Structure Forces and Electrostatic Modulation of
Biomolecular Interactions in Solution; J. Phys. Chem. B 111, 227 (2007)
[7] S A Hassan and E L Mehler, A Critical Analysis of Continuum Electrostatics:
The screened Coulomb potential-implicit solvent model and the study of
the alanine dipeptide and discrimination of misfolded structures of
proteins; Proteins 47, 45 (2002)
[8] S A Hassan and E L Mehler, Modeling Aqueous Solvent Effect through Local
Properties of Water, in Modeling Solvent Environments: Applications to
Simulations of Biomolecules, Ed. M Feig, Wiley-VCH (2010)
Top
SCPISM example input file
---------------------------------------------------------------------
* energy, minimization and dynamics with the SCPISM
open read card unit 10 name top22.inp
read rtf unit 10 card
close unit 10
open read card unit 10 name par22.inp
read para unit 10 card
close unit 10
open read unit 10 card name SCPISM.inp
! define system
generate main setup
open read unit 2 card name filename.crd
read coor card unit 2
close unit 2
! set up all options for the run (nbond cutoff distance, shake, vdW, etc)
! these options can also be specified in the 'energy' command line, as usual
! request SCPISM
SCPI HYDR UISM 10
! this will activate electrostatics and a simple hydrophobic term;
! if other model is used to describe hydrophobic interactions then
! replace the above command line by: SCPI UISM 3
! now calculate energy, minimize structure and run dynamics
ener
mini [options]
dyna [options]
! exit SCPISM
SCPI END
! if needed, the model can be requested again at any point
stop
-----------------------------------------------------------------------
SCPISM example input file
---------------------------------------------------------------------
* energy, minimization and dynamics with the SCPISM
open read card unit 10 name top22.inp
read rtf unit 10 card
close unit 10
open read card unit 10 name par22.inp
read para unit 10 card
close unit 10
open read unit 10 card name SCPISM.inp
! define system
generate main setup
open read unit 2 card name filename.crd
read coor card unit 2
close unit 2
! set up all options for the run (nbond cutoff distance, shake, vdW, etc)
! these options can also be specified in the 'energy' command line, as usual
! request SCPISM
SCPI HYDR UISM 10
! this will activate electrostatics and a simple hydrophobic term;
! if other model is used to describe hydrophobic interactions then
! replace the above command line by: SCPI UISM 3
! now calculate energy, minimize structure and run dynamics
ener
mini [options]
dyna [options]
! exit SCPISM
SCPI END
! if needed, the model can be requested again at any point
stop
-----------------------------------------------------------------------