CHARMM Emprical Energy Function Parameters
This section describes parameters in the CHARMM empirical
energy function.
* Menu:
* Overview:: Overview of CHARMM parameter file by A. D. Mackerell Jr.
* Multiple:: Rules for the use of multiple dihedrals in CHARMM22
* Conversion:: Rules for conversion of old nucleic acid rtf and
param to CHARMM22 format
* PARMDATA:: Description of Parameter Files available for general use.
Overview of CHARMM parameter files
By Alexander D. MacKerell Jr., Aug. 1995
This section of the documenation contains a brief description
of the contents of a parameter file. The CHARMM parameter file
contains the information necessary to calculate energies etc. when
combined with the information from a PSF file for a structure.
Information on the keywords found in the parameter file is in IO.DOC.
(A) * CHARMM example parameter file
*
(B) BOND
H O 500.0 1.00
(C) ANGLe (THETa)
H O H 100.0 104.51 20.0 1.70
(D) DIHEdral (PHI)
HT CT CT HT 10.0 3 180.0
X CT CT X 10.0 3 180.0
(E) IMPH
O C CT N 5.0 1 0.0
X C CT X 5.0 1 0.0
X X CT N 5.0 1 0.0
O X X N 5.0 1 0.0
(F) NBONDed nonbond-spec
H 0.00 -0.046 0.2245 0.00 -0.023 0.2245
O 0.00 -0.120 1.8000 0.00 -0.060 1.8000
(G) NBFIX
H O -0.30 1.50 -0.15 1.50
(H) HBONDs hbond terms (IO.DOC)
H O -0.00 1.00
(I) END
The parameter file starts with a title (A) which contains
information on the origins and applicability of that file.
Section (B) BONDs, contains information on all bond force
constants and equilibrium geometries. In this as well as the
remainder of the parameter file the bonds etc. are specified by the
atom type associated with each IUPAC atom in the topology file.
Section (C) ANGLes or THETas, are specified by 3 atom types
followed by the force constant and equilibrium geometry. If a
Urey-Bradley term is desired between the 1 and 3 atom types of the
angle a second U-B force constant and equilibrium geometry are
included.
Section (D) DIHEdrals (PHI), contains the 4 atom types
specifing a dihedral followed by the force constant, the multiplicity
of the dihedral and the minimium geometry of the dihedral. With
dihedrals wildcards, X, as shown may be included for the terminal
atoms. Also, multiple dihedrals of different multiplicities may be
specified for a single dihedral as outlined below.
Improper dihedrals (E) IMPH, used for out of plane motions are
specified in the same fashion as dihedrals. The use of wildcards, X,
is also allowed in a number of variations. Multiple improper
dihedrals are not supported.
Parameters for (F) NONBonded VDW parameters may be specified
in two ways. Initially the Tanford-Kirkwood Formula was used where
the atom polarizabilities, Number of effective electrons, and (minimum
radius)/2 were required. In this formulation the first term following
the atom type is the atom polarizability, the second term is the
number of effective electrons and must be positive in order to specify
the Tanford-Kirkwood Formula and the third term is the (minimum
radius)/2. If the second term is negative, then the first number is
ignored, the second term is the well-depth (epsilon) and the third
term is the (minimum radius)/2. Both formulations use the
Lennard-Jones 6-12 formula to determine the VDW interactions, in the
first method the Tanford-Kirkwood Formula is used to calculate the
well-depth (epsilon) and in the second method it is used directly.
With both formulations a second set of 3 numbers may be specified to
indicate the VDW parameters to be used for the calculation of 1-4
nonbonded interactions. Wildcards (*, %, etc. see MISCOM.DOC) may be
used with the NONBond as well as the NBFIX and HBOND sections of the
parameter file.
The NBFIX section (G) allows VDW interactions between specific
atom pairs to be modified. This is done by specifing the 2 atom types
followed by the well depth and the minimum radius (not (minimum
radius)/2 as in NBOND). A second well depth and minimum radius may be
specified to determine the 1-4 interactions.
The final section (H) contains the hydogen bond well depths
and minimum radii for various atom pairs. In current versions of the
CHARMM parameter sets (PARAM19, CHARMM22 protein and nucleic acid
parameters) hydrogen bonding is included in the electrostatic and VDW
interactions. Thus, the HBOND well depth is set to -0.00 and in most
calculations IHBFRQ should be set to 0 to avoid updating the hydrogen
bond lists. This facility is still supported to allow calculations
using the Lennart Nilsson nucleic acid parameters, AMBER parameters
and for analysis of hydrogen bond geometries. It should be noted that
both the NBOND and HBOND keywords are followed by a number of keywords
dictating truncation schemes, 1-4 interaction treatments and
dielectric constants, amoung others. These specifications are of the
upmost importance for relabile calculations and deviations from the
default values supplied with the parameter files should be done with
the utmost caution.
Rules for the use of multiple dihedrals in CHARMM24
1) The association of 1 or more dihedrals with different
multiplicities to a specfic dihedral type (as specified
by atom types) is specified by the presence of 2 or
more dihedral parameters in the parameter file. When
multiple dihedrals are read in the parameter file CHARMM22
will list those dihedrals in the output file (Note: the following
type message "PARRDR> Multiple terms for dihedral type: INDEX 427
CODE31141959 CT3 -OS -CD -OB" indicates that the multiple
dihedral has been successfully read).
2) If dihedral angles are AUTOGENERATED, then the RTF should
not specify them again. Additional dihedrals in the RTF will
be ignored and warnings given.
3) Without AUTOGENERATE, each dihedral should appear only once
in the RTF. Multiple listings of a dihedral will be ignored
and warnings given.
4) The order of the dihedral entries associated with
a specific dihedral is important; they must be placed
sequentially in the parameter file. If they are not sequential
errors will be given. This is new in C24B1 and later versions.
For example:
P ON2 P2 ON2 0.03 2 0.0
P ON2 P2 ON2 0.03 3 0.0
will place both a 2-fold and a 3-fold term on the P-ON2-P2-ON2
dihedral.
5) Wildcards may be used in the parameter file to specify multiple
dihedrals(ie. X C1 C2 X), however, all the dihedrals in the
parameter file associated with that dihedral type must be
wildcards. Use of wildcards with multiple dihedrals is NOT
recommeded.
6) Specific dihedral entries always override wildcard entries.
For example:
X C2 C3 X 100.0 1 180.0
C1 C2 C3 C4 100.0 2 180.0
X C2 C3 X 100.0 3 180.0
will assign the 2-fold term to C1-C2-C3-C4 while 1-fold and
3-fold terms would be assigned to C5-C2-C3-C6 and any other
dihedral centered about the C2-C3 bond. This assignment of
the multiple terms to a number of dihedrals is why the use
wildcards for the specification of multiple dihedrals in NOT
recommeded. The preferred method is as follows:
X C2 C3 X 100.0 2 180.0
C5 C2 C3 C6 100.0 1 180.0
C5 C2 C3 C6 100.0 3 180.0
will assign the 1-fold and 3-fold terms to C5-C2-C3-C6 and the
2-fold term to C1-C2-C3-C4 and any other dihedral centered about
the C2-C3 bond. This limits the potential for multiple dihedrals
being mistakenly assigned to a dihedral centered on the C2-C3 bond.
Thus, it is advised that when creating a multiple dihedral all
4 atom types be explicitly stated and, if necessary, new atom
types be created to avoid conflicts.
7) This design is such that previous CHARMM topology and parameter
files for proteins are compatible with CHARMM24. However, due
to complexities in the multiple dihedral setup for the nucleic
acid sugars (ribose and deoxyribose) the nucleic acid topology
and parameter files are NOT compatible with CHARMM22. In order
to make them compatible the following alterations must be
performed. Alternatively, the altered files may be obained
from Alexander D. MacKerell Jr.
Rules for conversion of old nucleic acid rtf and param to CHARMM22 format
The following conversion rules apply to CHARMM22. Compatability with
C24B1 and later versions will be insured if the multiple dihedrals in
the converted parameters are sequential, as disscused above.
ALL-HYDROGEN
Protocol for conversion of all-hydrogen nucleic acid topology and
parameter files (topnah*.inp and parnah*.inp) from a CHARMM21 or
previous format to a format compatible with CHARMM22. This change is
due to a new methodology for the treatment of multiple dihedrals in
CHARMM22.
In Topology File (TOPNAH1.INP, TOPNAH1E.INP, TOPNAH1R.INP)
1) Create a new atom type, OSS
2) Convert the atom type of all O4' atoms to OSS
In Parameter File (PARNAH1.INP)
1) Copy all OS parameters (bonds, angles, dihedrals etc.)
and in the copy change OS to OSS. Be sure that the
original OS parameter remains. Some OS to OSS copies
can be avoided (such as OS P terms), however, one
must be careful that all the necessary OSS parameters
relating to O4' are present. Creating extra OSS
parameters which are unused is not a problem. One
exception occurs with the dihedral OS CH CH OS, where
only one of the terminal OS atom should be converted
to OSS.
2) In the DIHEDRAL (PHI) parameters under the heading
"WILMA OLSON SUGAR MODEL" the following steps must be
performed once all the OSS dihedral parameters are
created.
A) In all the explicit OS terms which don't include
wildcards (X) or P atom types and have both 2 and
3-fold periodicities (2nd of 3 numbers following the
dihedral) the 2nd 3-fold term must be commented out
with a !.
B) Of the new explicit OSS terms the following
3-fold terms must be commented out with a !.
OSS CH CH OS 1.4000 3 0.0000
OH CH CH OSS 1.4000 3 0.0000
Lastly, when generating the structure be sure only the AUTOGENERATE
ANGLE term is used. (i.e. do NOT use AUTOGENERATE DIHEDRAL).
At this point the topology and parameter files should be
compatible with CHARMM22 (but not CHARMM21 or a previous version of
CHARMM). A test should be performed on a (deoxy)ribose containing
containing compound. In this test the energies should be calculated
1) using CHARMM21 or a previous version using the original, unmodified
topology and parameter files and 2) with CHARMM22 using the modified
OSS containing topology and parameter files. These energies should be
equivalent.
EXTENDED (UNITED) ATOM
Protocal for the conversion of extended (united,explicit) atom
nucleic acid topology and parameter files from CHARMM21 or previous
format to a format compatible with CHARMM22. This change is due to a
new methodology for the treatment of multiple dihedrals in CHARMM22.
In Topology File (TOPRNA10 or TOPRNA10R)
1) Create 2 new atom types, OSS and OST
2) Convert the atom type of all O4' atoms to OSS except
in the the patch PRES DEOX where it must be changed
to atom type OST. This conversion to OST must also be
performed in any residue, such as RESI DRIB, in which
deoxyribose is used explicitly.
3) In the patch PRES DEOX add the line:
ATOM O4' OST -0.30 ! (check the charge)
before the GROUP statement and comment out the terms
!DELETE DIHE O4' C4' C3' O3' ! WE NEED THIS AS A MULTIPLE TERM IN DEOXY
!DIHE O4' C4' C3' O3' ! threefold
!DIHE O4' C4' C3' O3' ! twofold
such that no alterations in the dihedral setup are made.
In Parameter File (PARDNA10.INP)
1) Copy all OS parameters twice (bonds, angles, dihedrals etc.);
in the first copy change OS to OSS and in the second change OS
to OST. Be sure that the original OS parameter remains. Some
OS to OSS(OST) copies can be avoided (such as terms in which OS
is adjacent to P), however, one must be careful that all the
necessary OSS(OST) parameters relating to O4' are present.
Creating extra OSS(OST) parameters which are unused is not a
problem. One exception occurs with the dihedral OS CH CH OS,
where only one of the terminal OS atom should be converted
to OSS(OST).
2) In the DIHEDRAL (PHI) parameters under the heading
"WILMA OLSON SUGAR MODEL" the following steps must be
performed once all the OSS(OST) dihedral parameters are
created.
A) In all the explicit OS terms which don't include
wildcards (X) or P atom types and have both 2 and
3-fold periodicities (2nd of 3 numbers following the
dihedral) the 2nd term must be commented out
with a ! (mostly 3-fold terms and 1 or 2 2-fold term).
B) Of the new explicit OSS terms the following
3-fold terms must be commented out with a !.
OSS CH CH OS 1.4000 3 0.0000
OH CH CH OSS 1.4000 3 0.0000
C) Maintain all of the OST dihedral terms.
An example of the additions/alterations to pardna10.inp are
listed below.
BOND
HO OSS 450.0000 0.9600
HO OST 450.0000 0.9600
OSS CH 292.0000 1.4300
OSS C2 292.0000 1.4300
OST CH 292.0000 1.4300
OST C2 292.0000 1.4300
C3 OSS 292.0000 1.38
C3 OST 292.0000 1.38
C OSS 292.0000 1.43
C OST 292.0000 1.43
THETA
OSS C2 C3 150.5000 111.0000
OSS C2 CH 70.0000 112.0000
OSS C2 C2 82.0000 112.0000
OST C2 C3 150.5000 111.0000
OST C2 CH 70.0000 112.0000
OST C2 C2 82.0000 112.0000
C2 CH OSS 46.5000 111.0000
C2 CH OST 46.5000 111.0000
C3 CH OSS 46.5000 111.0000
C3 CH OST 46.5000 111.0000
CH CH OSS 46.5000 111.0000
CH CH OST 46.5000 111.0000
OSS CH NS 46.5000 111.0000
OSS CH NH2E 46.5000 111.0000
OST CH NS 46.5000 111.0000
OST CH NH2E 46.5000 111.0000
C2 OSS C2 82.0000 111.5000
CH OSS CH 46.5000 111.5000
HO OSS CH 46.5000 107.3000
HO OSS C2 46.5000 107.3000
C2 OST C2 82.0000 111.5000
CH OST CH 46.5000 111.5000
HO OST CH 46.5000 107.3000
HO OST C2 46.5000 107.3000
CH OSS C3 46.5 107.3
CH OST C3 46.5 107.3
C OSS C3 46.5 120.5
C OST C3 46.5 120.5
O C OSS 70.0 120.0
O C OST 70.0 120.0
CH C OSS 70.0 125.3
NA C OSS 70.0 120.0
CH C OST 70.0 125.3
NA C OST 70.0 120.0
OSS CH CS 46.5 111.0
OST CH CS 46.5 111.0
PHI
X CH OSS X 0.9000 3 0.0000
X CH OST X 0.9000 3 0.0000
X C2 OSS X 0.5000 3 0.0000
X C2 OST X 0.5000 3 0.0000
! OSS SUGAR TERMS
OSS CH CH OS 0.5000 2 0.0000
!OSS CH CH OS 1.4000 3 0.0000 Should be commented out
OH CH CH OSS 0.5000 2 0.0000
!OH CH CH OSS 1.4000 3 0.0000 Should be commented out
OSS CH CH CH 0.5000 2 0.0000
OSS CH CH CH 1.4000 3 0.0000
OSS CH C2 CH 1.0000 2 0.0000
OSS CH C2 CH 1.4000 3 0.0000
OSS CH CH C2 1.4000 3 0.0000
OSS CH CH C2 0.5000 2 0.0000
OSS C2 C2 C2 1.4 3 0.0
OSS C2 C2 C2 0.5 2 0.0
! OST SUGAR TERMS
OST CH CH OS 0.5000 2 0.0000
OST CH CH OS 1.4000 3 0.0000
OH CH CH OST 0.5000 2 0.0000
OH CH CH OST 1.4000 3 0.0000
OST CH CH CH 0.5000 2 0.0000
OST CH CH CH 1.4000 3 0.0000
OST CH C2 CH 1.0000 2 0.0000
OST CH C2 CH 1.4000 3 0.0000
OST CH CH C2 1.4000 3 0.0000
OST CH CH C2 0.5000 2 0.0000
OST C2 C2 C2 1.4 3 0.0
OST C2 C2 C2 0.5 2 0.0
! additional terms for tRNA
OSS CH CS CF 1.5 3 0.0
OST CH CS CF 1.5 3 0.0
C2 CH C OSS 1.5 3 0.0
C2 CH C OST 1.5 3 0.0
X C OSS X 1.8 2 180.00
X C OST X 1.8 2 180.00
! THE FOLLOWING TERMS UNDER THE HEADER
! "WILMA OLSON SUGAR MODEL":
! SHOULD BE COMMENTED OUT
!OS CH CH OS 1.4000 3 0.0000
!OS CH CH CH 1.4000 3 0.0000
!OH CH CH OS 1.4000 3 0.0000
!OS CH C2 CH 1.4000 3 0.0000
!OS CH CH C2 0.5000 2 0.0000
!OS C2 C2 C2 0.5 2 0.0
IMPHI
OSS X X CH 31.5000 0 35.2600
OST X X CH 31.5000 0 35.2600
CH OSS C2 NS 31.5000 0 35.2600
CH OSS CH NS 31.5000 0 35.2600
CH OSS C2 NH2E 31.5000 0 35.2600
CH OSS CH NH2E 31.5000 0 35.2600
CH OST C2 NS 31.5000 0 35.2600
CH OST CH NS 31.5000 0 35.2600
CH OST C2 NH2E 31.5000 0 35.2600
CH OST CH NH2E 31.5000 0 35.2600
NBONDED
OSS 0.64 7.0 1.6
OST 0.64 7.0 1.6
Lastly, when generating the structure be sure only the
AUTOGENERATE ANGLE term is used. (i.e. do NOT use AUTOGENERATE
DIHEDRAL).
At this point the topology and parameter files should be
compatible with CHARMM22 (but not CHARMM21 or a previous version of
CHARMM). A test should be performed on a (deoxy)ribose containing
containing compound. In this test the energies should be calculated
1) using CHARMM21 or a previous version using the original, unmodified
topology and parameter files and 2) with CHARMM22 using the modified
OSS containing topology and parameter files. These energies should be
equivalent.
(A) Topology files
(1) top_all22_lipid.inp all hydrogen RTF for lipids
(2) top_all22_prot.inp all hydrogen RTF for proteins
(3) top_all22_na.inp all hydrogen RTF for nucleic acids
(4) top_all22_prot_na.inp all hydrogen RTF for proteins and nucleic acids
(5) top_all22_model.inp all hydrogen RTF for protein model cmpds
(6) toph19.inp extended atom RTF for proteins
(7) toprna10r_22.inp extended atom RTF for nucleic acids
(B) Parameter files
(1) par_all22_lipid.inp all hydrogen parameters for lipids
(2) par_all22_prot.inp all hydrogen parameters for proteins
(3) par_all22_na.inp all hydrogen parameters for nucleic acids
(5) par_all22_prot_na.inp all hydrogen parameters for proteins and nucleic acids
(6) param19.inp extended atom parameters for proteins
(7) pardna10_22.inp extended atom parameters for nucleic acids
The charmm22 all-hydrogen topology and parameter sets may be
considered to be stable, however, further changes cannot be excluded.
The bulk of the changes are expected to be additions leading to an
expanding set of parameters which are compatible across proteins,
nucleic acids, lipids, and, ultimately, carbohydrates. The
carbohydrate parameter work is still in progress by John Brady and
coworkers. See the file toppar_all.history for a listing changes in
the files over time. top_all22_model.inp includes the majority of
model compounds used in the protein parameterization and is to be used
in conjunction with par_all22_prot.inp. top_all22_prot_na.inp and
par_all22_prot_na.inp contain both the protein and nucleic acid
parameters allowing for calculations on protein-nucleic acid
complexes. The lipid parameters may be combined in a similar fashion,
although it hasn't been performed at present. The extended atom
parameters for proteins are the same as those included with CHARMM20
which are based on Wally Reiher's thesis. For the extended atom
nucleic acid parameters those of Nilsson and Karplus, J. Comp. Chem.
7:591-616, 1986 are used which were also included in the CHARMM20
release and are the only set to include explicit hydrogen bonding
terms. Some alterations of the extended atom nucleic acid topology
and parameter files have been made in order to maintain compatibility
with the multiple dihedral scheme in CHARMM22. Please send all
remarks and suggestions to alex@mmiris.ab.umd.edu. ADM Jr., July, 1995
Suggested references for the CHARMM22 all-hydrogen sets. Please
contact ADM Jr., for updates.
For the proteins
MacKerell Jr., A.D. and Karplus, M. All-hydrogen empirical potential
for molecular modeling and dynamics studies of proteins using the
CHARMM22 force field, Manuscript in preparation.
and
MacKerell Jr., A.D., Bashford, D., Bellott, M., Dunbrack Jr., R.L.,
Field, M.J., Fischer, S., Gao, J., Guo, H., Ha, S., Joseph, D.,
Kuchnir, L., Kuczera, K., Lau, F.T.K., Mattos, C., Michnick, S., Ngo,
T., Nguyen, D.T., Prodhom, B., Roux, B., Schlenkrich, M., Smith.,
J.C., Stote, R., Straub, J. Wiorkiewicz-Kuczera, J. and Karplus, M.,
Self-consistent parameterization of biomolecules for molecular
modeling and condensed phase simulations. FASEB Journal 1992, 6:A143.
For the nucleic acids
MacKerell Jr., A.D., Wiorkiewicz-Kuczera, J. and Karplus, M. An
all-atom empirical energy function for the simulation of nucleic
acids, In Press.
for the lipids
Schlenkrich, M., Brickmann, J., MacKerell, A.D., Jr., and Karplus, M.
Empirical Potential Energy Function for Phospholipids: Criteria for
Parameter Optimization and Applications, in "Membrane Structure and
Dynamics," K.M. Merz and B. Roux, Eds. Birkhauser, Boston, To be
published for the 1996 Biophysical Society Meeting.
NIH/DCRT/Laboratory for Structural Biology
FDA/CBER/OVRR Biophysics Laboratory