Re: RATTLE Error During Minimization of Charmm Gui Generated Bilayer

From: Chitrak Gupta (
Date: Tue Dec 15 2015 - 12:46:36 CST

Hi Begum,

I have never used charmm27, but I have used charmm36 and have ran into this
problem quite a few times. Here are my questions/suggestions:

1. How are you merging your lipid with your peptide? In my experience,
TopoTools works a lot better than using psfgen

2. Make sure your merged PDB doesn't have anything funny. For example, if
some atom has its X,Y,Z coordinates as 0.000,0.000,0.000, it basically
means coordinates were not set for that atom. Look through your PDB (maybe
generate a script to check it) whether you have such atoms.

3. Also look at the number of waters. When charmm-gui generates a psf/pdb,
it assigns all the waters into one segment. Problem with psfgen is that it
cannot handle more than 9999 waters within a segment. So, when you take a
charmm-gui generated PDB and modify it using psfgen, your water numbers are
completely messed up. You now have multiple waters with the same number.
You will need to re-segment your waters, keeping 9999 waters in each

Hope this helps.


On Tue, Dec 15, 2015 at 11:27 AM, begüm alaybeyoğlu <>

> Dear NAMD users,
> I am having trouble with the Charmm Gui generated POPE bilayer. After
> having run all the minimization-equilibration-production steps, I ran a 25
> ns production simulation (step.7.1) without any problems. What I wanted to
> do was to get the last snapshot of my equilibrated bilayer, place my
> peptide on top of it, autoionize the system and run a quick minimization &
> equilibration before I start the peptide translocation simulation.
> Now I keep getting the RATTLE error (for atoms of bilayer always) just
> after the minimization is completed and I tried all of the previously
> suggested solutions, such as decreasing the timestep or nonbondedFreq or
> fullElectFrequency or stepspercycle and increasing the minimization steps
> etc. When I plotted the energy I saw that around 2500 steps of minimization
> was enough already. I tried slow heating of the system, but the error did
> not change. By decreasing my dcdfreq to 10 steps, I observed that H atoms
> (bound to C or N) actually start to overlap during the minimization, so the
> cause of the error is probably this.
> I have run many similar simulations with charmm27 before without any
> problems.This is the first time I built the whole system from scratch
> using the charmm36 parameters, so I am not sure what is going on. Any
> suggestions are welcome.
> Thanks.
> Begum Alaybeyoglu

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:22:20 CST