Can't run NAMD with cellulose-builder structure

[Jaqueline Sousa]

Dear developers,

My pdb and psf files won't work with NAMD and CHARMM36 parameters. I'm using par_all36_carb.prm downloaded from the McKarell Lab website, the first problem I had was that the ions and water parameters wouldn't work in the minimization, and the structure invariable broke. So I downloaded separate str files for water and ions in McKarell lab website, they seemed to work nicely at first, but whenever I run any kind of heating (even at very small temperatures)  or production the energy of the system and the atom velocities get too high. Have you guys any idea of what must be going on?

Att,

Jaqueline Sousa

[Thiago Gomes]

Dear Jaqueline,

thanks for reporting that.
Could you please send me pdb and psf produced by cellulose-builder so I could inspect them?

A few guesses:

(just guesses though. I can't be sure without inspecting your input files for NAMD - solvated pdb and psf , equilibration and heating/production procedure and NAMD output files):

1) depending on the size of the crystallite you may need 1000+ minimization steps before running heating and production. Plotting potential energy versus time is a useful manner to check if you need further minimization steps before heating.

One should find useful to perform minimization until the total potential energy of the system is more negative than its mean total potential energy will be after the first velocity vectors are distributed and simulation is run for a while (i.e. the first temperature point in the heating procedure).

Otherwise, the system may still have some local strain that may result in velocities getting too high upon start simulating at even very low temperatures, say, 4 kelvin. Since you don't know how much is the mean total potential energy at, say, 4 K, guess an exaggerated number of minimization steps. After you get rid of the "atom velocities got too high" error you may want to readjust the number of minimization steps for future simulations on similar systems.

2) check for misplaced water or anions in the middle of crystallites and also keep an eye on the water density. (Also, see packmol for solvating and for general mixtures).

3) using periodic boundary conditions might be tricky (check your basevectors), especially when trying to use a barostat along with periodic covalent bonds along the cellulose chains' direction. As soon as you get an "equilibrium" average volume in the NPT ensemble, switch to the NVT. You will gain performance anyways by avoiding the barostat.

Best wishes,

Thiago

--
You received this message because you are subscribed to the Google Groups "cellulose-builder" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cellulose-buil...@googlegroups.com.
To post to this group, send email to cellulos...@googlegroups.com.
Visit this group at http://groups.google.com/group/cellulose-builder.
For more options, visit https://groups.google.com/d/optout.

[Jaqueline Sousa]

Hi Thiago,

Thank you, I'm sending both psf and pdb files. I didn't use pbc conditions to build the crystal, and my structure isn't that big (~40000 atoms). I'll try minimizing it for more steps as you suggested.

https://drive.google.com/file/d/0B_DYsLFK3aMbWHFhZ3dnMUtmclU/view?usp=sharing
https://drive.google.com/file/d/0B_DYsLFK3aMbOHZ0VllmVkJodFk/view?usp=sharing

Att,

Jaqueline Sousa