Setting up membrane protein simulation from existing conformation?

From: Per Larsson (
Date: Wed May 07 2014 - 14:17:41 CDT

Dear Namd users

Apologies of this is a very basic question. I am only starting out with name, having much previous Gromacs experience, but I need to test some of the accelerated md protocols available in namd.

I am having problems generating a complete psf for my system, which contains one protein, a popc bilayer and water molecules, everything properly ordered.

I can generate the psf for the protein and membrane, but have run into problems when dealing with the solvent. I am using namd 2.9 and the charmm36 forcefield, recently downloaded from the MacKerell website.

Here is my complete script to psfgen:

# Use the specified CHARMM27 topology file.
topology toppar/top_all36_prot.rtf

# create an alias for TIP3
pdbalias residue SOL TIP3

# Build three segments, one for each chain.
segment solv {
  auto none
  pdb solvent.pdb

# Load the coordinates for each segment, and use alias for atom names in the input file
pdbalias atom TIP3 OW OH2
pdbalias atom TIP3 HW1 H1
pdbalias atom TIP3 HW2 H2
coordpdb solvent.pdb solv

# Write out the psf file
writepsf solvent.psf

Running this in namd 2.9 gives the following errors:

---- *
psfgen) Created by CHARMM version 36 1
psfgen) aliasing residue SOL to TIP3
psfgen) building segment SOLV
psfgen) disabling angle autogeneration
psfgen) disabling dihedral autogeneration
psfgen) reading residues from pdb file foo.pdb
psfgen) unknown residue type TIP3
psfgen) extracted 31 residues from pdb file
psfgen) Info: generating structure...psfgen) unknown residue type TIP3
ERROR: failed on end of segment
MOLECULE DESTROYED BY FATAL ERROR! Use resetpsf to start over.

Now, this sort of makes sense in that indeed there is no TIP3 residue defined in the toppar/top_all36_prot.rtf file, but raises the question where in the c36-files this is defined precisely? I have seen tutorial material that groups the psfgen of protein and solvent together, and in which they use top_all22_prot.rtf, but that does not help me either, unfortunately.

My other question is that I have seen hints trying to google my way here that there are potentially issues with feeding pdb-files with large (I have > 100000) numbers of water molecules to psfgen. Is this true, and in this case is the only other option to use the solvate program?

Sorry for the long mail, but hopefully I can move forward with this fairly quickly with some help from others.

/Per Larsson

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:20:46 CST