[AMBER] vlimit exceeded and volume of ucell too big, too many subcells during adQM/MM dynamics

From: Ruth Helena Tichauer <rhtichau.laas.fr>
Date: Thu, 11 Jul 2019 09:36:30 +0200

Dear Amber experts,

I’m starting a new thread of the same previously reported issue but changing the subject to add clarity to it.

For comparison purposes with a previous study, I’m attempting to run adaptive solvent QM/MM dynamics of a system consisting in a protein-ligand complex to which a loop of another protein is docked and kept in this position by restraints applied to its borders. A Mg2+ ion is also present in the active site and I wish to treat quantum mechanically water molecules in the vicinity of the ligand (GTP). To do so, this is the input file I used:

Constant Temp 300K adQM/MD simulation
 irest=1, ntx=5,
 ntb=2, pres0=1.0, ntp=1, taup=1.0,
 ntt=3, gamma_ln=2.0, ig=-1,
 nstlim=50000, dt=0.001,
 ntpr=10, ntwx=10, ntwr=100, ioutfm=1,
 ntc=2, ntf=2,
 noshakemask = '.176-189,474-494,521-534,900-909,917-933,2646-2689,2828-2851,2941-2946',
 restraintmask = ':170,185,.CA,C,N',

From the 4 partitions that result in mdout files 000, 001, 002 and 003, mdcrd and restrt files are created for the first one (000) only. Is this behaviour normal?

Aside, the simulation is stopped due to the fourth partition which mdout file displays the following error message:

SANDER BOMB in subroutine nonbond_list
 volume of ucell too big, too many subcells
 list grid memory needs to be reallocated, restart sander

Yet, when checking the size of the QM region through the .pdb files written thanks to the “writepdb” keyword, this region doesn’t ‘blow up’ (see attachments named qmmm_region.pdb.*). Nevertheless, when checking this same region through the .xyz files generated thanks to the “print_qm_coords” keyword (QM_coords.* files), a residue from the loop of the second protein appears to be far away, as when visualising a wrapped trajectory that hasn’t been unwrapped. Moreover, in the .xyz file corresponding to the partition that leads to the halt of the simulation (QM_coords.003.xyz <http://qm_coords.003.xyz/><http://qm_coords.xyz/ <http://qm_coords.xyz/>>), a water molecule is observed with the “wrapped” residue from the second protein such that this water molecule appears in both sides i.e. on the side of the protein-ligand complex and on the side of the second protein loop. Hence, I suspect this water molecule on “both sides” could be leading to the repeated "vlimit exceeded” messages I get before the crash:

vlimit exceeded for step 156; vmax = 20.4418
wrapping first mol.: -45.29463 -64.05628 0.00000
vlimit exceeded for step 157; vmax = 20.3292
wrapping first mol.: -45.54621 -64.41207 0.00000
vlimit exceeded for step 158; vmax = 20.2651
wrapping first mol.: -45.75854 -64.71234 0.00000

Could this be the case?

Any thoughts on what shall be done to pursue this adQM/MM simulation?
Thanking you in advance for any insights on this matter,



P.S: I’m using Amber 16 on a cluster external to the lab
AMBER mailing list
Received on Thu Jul 11 2019 - 01:00:02 PDT
Custom Search