A PSF file, also called a protein structure file, contains all of the molecule-specific information needed to apply a particular force field to a molecular system. The CHARMM force field is divided into a topology file, which is needed to generate the PSF file, and a parameter file, which supplies specific numerical values for the generic CHARMM potential function. The topology file defines the atom types used in the force field; the atom names, types, bonds, and partial charges of each residue type; and any patches necessary to link or otherwise mutate these basic residues. The parameter file provides a mapping between bonded and nonbonded interactions involving the various combinations of atom types found in the topology file and specific spring constants and similar parameters for all of the bond, angle, dihedral, improper, and van der Waals terms in the CHARMM potential function.
The PSF file contains five main sections of interest: atoms, bonds, angles, dihedrals, and impropers (dihedral force terms used to maintain planarity). The following is taken from a PSF file for ubiquitin. First is the title and atom records:
PSF 1 !NTITLE REMARKS original generated structure x-plor psf file 1231 !NATOM 1 PROT 1 MET N NH3 -0.300000 14.0070 0 2 PROT 1 MET HT1 HC 0.330000 1.0080 0 3 PROT 1 MET HT2 HC 0.330000 1.0080 0 4 PROT 1 MET HT3 HC 0.330000 1.0080 0 5 PROT 1 MET CA CT1 0.210000 12.0110 0 6 PROT 1 MET HA HB 0.100000 1.0080 0 7 PROT 1 MET CB CT2 -0.180000 12.0110 0
The fields in the atom section are atom ID, segment name, residue ID, residue name, atom name, atom type, charge, mass, and an unused 0. PSF files may be in either CHARMM or X-PLOR format, with the CHARMM format using an integer rather than a name for the atom type. NAMD requires the X-PLOR format, which is also more flexible since it is not tied to the specific order of atom types in a single parameter file. NAMD and VMD require that the order of atoms in any PDB, DCD, or other atomic data file exactly matches the order found in the PSF file.
Notice that hydrogen atoms are included, that multiple atoms in a residue may share the same type (e.g., HT1, HT2, and HT3 are of type HC), and that atoms of the same element may have different types (e.g., CA and CB, HA, and HT1). Much of the information in the PDB is also included in the PSF file. Often, the only information from the PDB file used by NAMD is the atomic coordinates. In some cases the contents of the occupancy and beta fields of the PDB file are used as parameters to optional simulation methods.
The covalent bond section lists four pairs of atoms per line:
1237 !NBOND: bonds 1 5 2 1 3 1 4 1 5 6 7 5 7 8 7 9 10 7 10 11 10 12 13 10
The angle section lists three triples of atoms per line:
2257 !NTHETA: angles 1 5 6 1 5 18 2 1 5 2 1 4 2 1 3 3 1 5 3 1 4 4 1 5 5 18 19
The dihedral and improper sections list two quadruples of atoms per line:
3293 !NPHI: dihedrals 1 5 7 10 1 5 7 8 1 5 7 9 1 5 18 20 1 5 18 19 2 1 5 7
204 !NIMPHI: impropers 18 5 20 19 20 18 22 21 30 32 27 31 30 27 32 31 32 30 33 34 32 30 34 33
The order of the atoms within each bond is not significant. The order of atoms within angles, dihedrals, and impropers is significant, but the order can be reversed without affecting the resulting potential. In no case is the relative order of different bonds etc. significant. Since every bond and improper is explicity listed in the topology file, the atom order within these terms tend to match the original order given. Angles and dihedrals are typically generated automatically and therefore appear in a sorted order.
There is a difference between X-PLOR formatted CHARMM PSF files, which VMD generates with the psfgen module and NAMD uses in conjunction with CHARMM parameter files, and PSF files generated by X-PLOR, which NAMD uses in conjunction with X-PLOR parameter files. Dihedral terms sometimes require multiple sinusoids to represent a torsional potential and therefore multiple parameter sets appear in the parameter file. In PSF files generated by X-PLOR multiple dihedrals would be indicated by duplicate dihedrals in the PSF file. When using CHARMM PSF and parameter files NAMD extracts any multiple dihedral terms directly from the parameter file and each dihedral appears only once in the PSF file.