PSF file format

[Don Bashford]

I'm more of a CHARMM person than a NAMD person, but these two programs
have the PSF in common, so I can answer your questions.

You are correct about the electric charges, as well as the other
guesses you made about the PSF snippet you posted.  

"Donors" and "acceptors" refer to hydrogen bonds.  Some of the older
molecular mechanics empirical functions used explicit H-bonding terms,
but in the 80s there was a move toward dropping these in favor charges
and van der Waals terms with parameters tuned to reproduce H-bonding
effects.  Even so, many molecular mechanics programs retain the
capability of explicit hydrogen bonding for purposes like counting up
H-bonds.

PSF stands for Protein Structure File.  You seem to be pretty clear
about the sort of information it contains, so I won't explain that,
but your guess about how it is generated is probably not correct.
The CHARMM program (from the Karplus group at Harvard Chem. Dept.)
generates a PSF after first reading a Residue Topology File (RTF) and
then the protein sequence.  The RTF has bond-connectivity and charge
information for all the amino-acid residue types, as well as
information about how conntectivity and so forth change when you link
them together into a peptide chain or make other covalent
modifications like a Cys-Cys crosslink or protonation of a
carboxylate.  This information, in combination with the desired
residue sequence, and perhaps some directives to make other covalent
modifications, is sufficient to allow construction of the PSF without
reading any coordinates.  The X-PLOR program (from Axel Brunger of
Yale) which is closely related to CHARMM also generates PSF files, but
in a slightly different format than CHARMM.

Some molecular mechanics programs don't generate their own PSFs
because it is a complicated thing to program.  Instead, they rely on a
program like CHARMM or X-PLOR to do it for them.  I don't know whether
NAMD is one of these, but I think that some of the older programs from
the Schulten group (NAMD's origin) did use externally generated PSFs.

Donald Bashford
The Scripps Research Institute