From: Tristan Croll (tristan.croll_at_qut.edu.au)
Date: Wed Sep 03 2014 - 02:26:27 CDT
You should always try to cc in the namd-l list to your replies.
As for the ideal number of steps, how long is a piece of string? There is no one answer to that – every system is different, and no amount of energy minimisation (or equilibration, unless you have access to some astonishing resources) will correct a structure that is fundamentally misfolded. The best it can do is give you a structure that will run without crashing. If you’re working from a homology model, I’d suggest doing at least some basic quality checks before going forward – have a look the packing, look for major clashes and for other signs of trouble (like hydrophilic residues poking into the core of the protein).
Hope this helps.
From: Sandhyaa Subramanian [mailto:sandysubram_at_yahoo.com]
Sent: Wednesday, 3 September 2014 2:07 PM
To: Tristan Croll
Subject: Re: namd-l: FATAL ERROR: Low global exclusion count! (9231 vs 9270) System unstable or pairlistdist or cutoff too small.
Thanks a lot Tristan Croll. Yes I did realize that I should have minimized the structure before starting the main protocol as mine is a theoretical protein. Now that I have minimized I should hopefully not face any problem.
I have another question (since I am fairly new to the field of MD simulations) - What would be the ideal number of steps for minimisation, equilibration and MD run esp. for a theoretical protein just obtained by homology modelling?
Thank you once again.
On Wednesday, 3 September 2014 3:57 AM, Tristan Croll <tristan.croll_at_qut.edu.au> wrote:
Sounds like there’s a geometric problem (atoms overlapping, an overlong bond, an angle way off from norm...) somewhere in your protein, which will cause a crash like this on startup. This is why you should always perform a second minimization with the protein atoms free before starting full dynamics. A fairly typical protocol is:
- Minimize for a few thousand steps (25,000 is usually overkill) followed by ~20,000 steps of MD, all with protein atoms either fixed or restrained. This allows the water to thoroughly “soak” the protein before you start it moving.
- Minimize for another few thousand steps with the protein free to move. This relaxes any “bad” geometrical features in the starting structure, so that it will start gracefully. Note that it’s always a good idea to check out any bad areas for yourself (running the structure past a server such as Molprobity will give you an exhaustive list) before you actually go ahead – it’s quite common for existing crystal structures to have quite serious errors which will badly bias your simulation.
- Finally, start your MD run. Some people choose to run a protocol of gradually heating the system up from 0 to 300K – personally, I don’t find that makes much of a difference in most cases.
Hope this helps.
From: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] On Behalf Of Sandhyaa Subramanian
Sent: Sunday, 31 August 2014 12:06 AM
Subject: namd-l: FATAL ERROR: Low global exclusion count! (9231 vs 9270) System unstable or pairlistdist or cutoff too small.
I encountered the following error while trying to equilibrate my system which is a protein solvated with 7A water box and minimized with 25,000 steps -
FATAL ERROR: Low global exclusion count! (9231 vs 9270) System unstable or pairlistdist or cutoff too small.
I tried changing the pairlistdist to 15.0 and the cutoff to 13.0, but in vain. I do not understand why the atom names are mentioned because when I checked the psf files they seem Ok to me. Also the protein was fixed during the minimization.
Please help at your earliest.
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:49 CST