This section covers the COMPASSv1 force field as it is implemented into
the towhee_ff_COMPASSv1 file in the ForceFields directory. All of the Towhee atom types for the COMPASSv1 force field
are listed, along with a short description of their meanings. For more
information about the COMPASSv1 force field see the Accelrys
web site. The COMPASSv1 force field is maintained by researchers at
Accelrys and only a portion of it has been published in the peer-reviewed
literature. I have only incorporated those parts that have been published.
Note that COMPASSv1 is a Lennard-Jones (9-6) force field and can only be
combined with other Lennard-Jones (9-6) force fields. I would like to
acknowledge David Rigby for providing very useful guidance about implementing
COMPASSv1. Any discrepencies (especially typos) from the published COMPASSv1
force field values are the sole responsibility of Marcus G. Martin, and
I welcome feedback on how this implementation compares with other programs.
References for COMPASSv1
The COMPASS family of forcefield is an ongoing subject of further parameterization.
The authors of the forcefield periodically revisit old parameters and refit them to
additional data. Towhee defines COMPASSv1 as the parameters published in the following
series of papers.
Typos and clarifications for COMPASSv1
With the large size of the COMPASSv1 force field it is not surprising that
there are a few typos in the published literature. Here I summarize and
clarify differences in my implementation from the literature.
COMPASSv1 in Towhee
- There are full van der Waals and coulombic interactions between
atoms connected by three bonds (1-4 interactions).
- Rigby et al. 1997: In the angle-angle-torsion term for ha-c4-c4-o2
the ijk/jkl/ijkl constant is set to 20.2006 while it should actually
be -20.2006 [personal communication between M.G. Martin and D. Rigby
August 09, 2001]
- Sun 1998: The mixing rules for epsilon are stated incorrectly.
The term in the denominator should read (ri0)6
+ (rj0)6. The + is missing in that
- Sun 1998: In the End Bond-Torsion for c4-c4-c4-c4 the K-L/I-J-K-L
k1 value is set to 0.0000 and I think it should be set to the same
value as the I-J/I-J-K-L k1 value of -0.0732. This correction was
made in Towhee.
The official force field name for COMPASSv1 in Towhee is 'COMPASSv1'. Here
I list all of the COMPASSv1 atom names for use in the towhee_input file,
along with a brief description taken from the COMPASSv1 papers. I have added
some comments where I thought clarification was needed, these are all
in [square brackets]. The COMPASSv1 naming convention has the first letter
as the element, the second character is the number of bonds to that element,
and the third character is a bit of a wildcard. Please note that the capitalization
and spacing pattern is important and must be followed exactly as listed
- 'ar' : argon
- 'c1o' : carbon in CO
- 'c2=' : carbon in CO2 and CS2
- 'c3a' : aromatic carbon [carbon with 3 bonds, aromatic]
- 'c4' : generic sp3 carbon [carbon with 4 bonds,
use only if none of the other c4 special cases are valid]
- 'c43' : sp3 carbon with three heavy atoms attached
- 'c44' : sp3 carbon with four heavy atoms attached
- 'c4o' : alpha carbon [sp3 carbon bonded to an
oxygen in an alcohol or ether]
- 'c4z' : carbon, sp3, bonded to -N3
- 'he' : helium
- 'h1' : nonpolar hydrogen [hydrogen bonded to carbon]
- 'h1h' : hydrogen in H2
- 'h1o' : strongly polar hydrogen [hydrogen bonded to oxygen]
- 'kr' : krypton
- 'ne' : neon
- 'n1n' : nitrogen in N2
- 'n1o' : nitrogen in NO
- 'n1z' : nitrogen, terminal atom in -N3
- 'n2=' : nitrogen [nitrogen with a single bond and a double
- 'n2o' : nitrogen in NO2
- 'n2z' : nitrogen, first atom in -N3
- 'n2t' : nitrogen, central atom in -N3
- 'n3o' : nitrogen in nitro group
- 'o1=' : oxygen in NO2 and SO2
- 'o1=*' : oxygen in CO2
- 'o12' : oxygen in nitro group (-NO2)
- 'o1c' : oxygen in CO
- 'o1n' : oxygen in NO
- 'o1o' : oxygen in O2
- 'o2' : generic oxygen with two bonds
- 'o2e' : ether oxygen
- 'o2h' : hydroxyl oxygen [alcohol oxygen]
- 'o2n' : oxygen in nitrates
- 'p4=' : phosphorous [phosphorous with three single bonds
and a double bond]
- 'si4' : generic silicon with four bonds attached
- 'si4c' : a subset of Si4, non-hydrogen atom attached
- 's1=' : sulfur in CS2
- 's2=' : sulfur in SO2
- 'xe' : xenon
COMPASSv1 uses point charges to represent the electrostatic interactions on molecules.
They have devised an automated system for assigning the point charges on each atom that
depends on the atom types, and the types of atoms bonded to each atom. This system is
implemented into Towhee Version 4.4.0 and later. In order to anable automatic charge assignment
you use inpstyle 2 in combination with charge_assignment 'bond increment'.
Please see the inpstyle 2 documentation for
further information on enabling this option.
Angle-Angle cross terms
COMPASSv1 is a class 2 force field and so you do need to include angle-angle
cross terms. However, I have automated the detection and type determination
for COMPASSv1 so I would recomend using the inpstyle = 2 feature in order
to determine these. The rest of the fancy cross terms for COMPASSv1 are
included in the bending angle and dihedral angle potentials.
Improper torsions are not automatically generated by the Towhee code as
the rules for determining where they are applied are not always straight-forward.
COMPASSv1 exclusively uses the out-of-plane version of the improper torsions,
and they are typically centered on an sp2 atom in order to
enforce planarity with its three neighbors. These torsions are listed
in the COMPASSv1 literature as i-j-k-l where the angle is the dihedral between
j-i-k and i-k-l, and the bonding pattern is i, k, and l all bonded to
atom j, and also not bonded to each other. In the towhee_input file this
out-of-plane improper torsion is listed only for atom j, and the atom
order there is i, k, l. Unfortunately, I have not yet set up the automation
for determining the improper types for COMPASSv1 yet as the impropers are
almost exclusively used on aromatics, and I have not yet tested the code
for those molecules. You will need to look in fftorsion.F in order to
find the types if you need the improper torsions for COMPASSv1.
Return to the Towhee Capabilities web page