[gmx-users] A charge group moved too... during backward transition of dual topologies

Nash, Anthony a.nash at ucl.ac.uk
Thu Aug 4 09:14:57 CEST 2016


I have tried equilibrating the non-posres simulation for an additional 2.5
ns, and taken just the final frame and velocity (so the timestep 5 5ns).
I’m also recorded a lot more to the output and I extended the fast
transition run from 100ps to 200ps so the change in softcore potentials
are not so extreme. Please note, this only happens in the reverse
transition of the dual topology, and only when I activate delta_lambda. I
end up seeing this:


Not all bonded interactions have been properly assigned to the domain
decomposition cells


A list of missing interactions:
               Angle of  14160 missing      3
         Proper Dih. of   9958 missing      3
               LJ-14 of   2214 missing      4
          exclusions of  91790 missing      8


Molecule type 'Protein2'
the first 10 missing interactions, except for exclusions:
         Proper Dih. atoms   40   43   49   52 global 63484 63487 63493
63496
               LJ-14 atoms   40   52           global 63484 63496
               Angle atoms   43   49   52      global 63487 63493 63496
         Proper Dih. atoms   44   43   49   52 global 63488 63487 63493
63496
               LJ-14 atoms   44   52           global 63488 63496
         Proper Dih. atoms   45   43   49   52 global 63489 63487 63493
63496
               LJ-14 atoms   45   52           global 63489 63496
               Angle atoms   50   49   52      global 63494 63493 63496
               Angle atoms   51   49   52      global 63495 63493 63496
               LJ-14 atoms   52   55           global 63496 63499


-------------------------------------------------------
Program mdrun_mpi_d, VERSION 5.0
Source code file: 
/home/sacat2/gromacs_source/gromacs-5.0/src/gromacs/mdlib/domdec_top.c,
line: 394


Fatal error:
18 of the 130046 bonded interactions could not be calculated because some
atoms involved moved further apart than the multi-body cut-off distance
(0.934022 nm) or the two-body cut-off distance (1.42 nm), see option -rdd,
for pairs and tabulated bonds also see option -ddcheck
For more information and tips for troubleshooting, please check the GROMACS
website at http://www.gromacs.org/Documentation/Errors
-------------------------------------------------------



Dr Anthony Nash
Department of Chemistry
University College London





On 03/08/2016 08:11, "gromacs.org_gmx-users-bounces at maillist.sys.kth.se on
behalf of Nash, Anthony"
<gromacs.org_gmx-users-bounces at maillist.sys.kth.se on behalf of
a.nash at ucl.ac.uk> wrote:

>Hi all,
>
>I¹m performing free energy calculations based on Crooks Fluctuation
>Theorem. To do this I¹ve used PMX to implement a dual topology. To keep
>things simple, it is a transmembrane polyleucine helical protein where one
>leucine is transforming into a serine - then there is the backward
>transition of serine into Leucine, required for the overlap of two
>gaussian distributions (forward and backward).
>
>I have a very complicated arrangement of steps to equilibrate the system,
>in very brief I have only included those involved once the protein is
>embedded in the bilayer (bilayer eq was around 40 ns over various stages).
>For a backward eq (serine->leucine) I used the parameters:
>
>free_energy = yes
>init_lambda = 1                 ;0=forward 1=backward
>delta_lambda = 0                ;equilibrium
>nstdhdl = 1
>sc-coul = yes
>sc-alpha - 0.3
>sc-sigma = 0.25
>sc-power = 1
>
>
>And the eq steps:
>
>-Steepest descent (<100 kjmol threshold)
>-NVT 100ps (carbon-alpha posres 1000 kjmol)
>-NPT 100ps (carbon-alpha posres 1000 kjmol - Berendsen pressure)
>-NPT 10ns (carbon-alpha posres 1000 kjmol - PR pressure)
>-NPT 2.5 ns (turn OFF posres)
>
>The above steps produce a system with serine in the polyleu. I then
>discard the first 1.5 ns and take 10 sets of coordinates and velocities
>from the last 1ns. This is for my Œfast¹ backward (serine->leucine) 100 ps
>transformation using the MDP parameters:
>
>free_energy = yes
>init_lambda = 1                 ;0=forward 1=backward
>delta_lambda = -0.00002         ;transitions
>nstdhdl = 1
>sc-coul = yes
>sc-alpha - 0.3
>sc-sigma = 0.25
>sc-power = 1
>
>
>When I run the 10 Œfast¹ transition production runs approximately half of
>them immediately result in the error:
>
>Step 5:
>The charge group starting at atom 63445 moved more than the distance
>allowed by the domain decomposition (1.420000) in direction Z
>distance out of cell 70.681579
>Old coordinates:  -48.465  256.312   81.655
>New coordinates:  -48.465  256.312   81.655
>Old cell boundaries in direction Z:    5.728   10.974
>New cell boundaries in direction Z:    5.728   10.974
>
>
>-------------------------------------------------------
>Program mdrun_mpi_d, VERSION 5.0
>Source code file: 
>/home/sacat2/gromacs_source/gromacs-5.0/src/gromacs/mdlib/domdec.c, line:
>4380
>
>
>Fatal error:
>A charge group moved too far between two domain decomposition steps
>This usually means that your system is not well equilibrated
>For more information and tips for troubleshooting, please check the
>GROMACS
>website at http://www.gromacs.org/Documentation/Errors
>-------------------------------------------------------
>
>
>I understand what this error means, but I was wondering whether I was
>doing something far more fundamentally wrong, given that every forward
>transition (same eq set up, just using init_lambda = 0 and delta_lambda =
>0.00002) never produce this error! They always work, and monitoring my
>work gives me an amazing gaussian distribution.
>
>I have put the 2.5ns production run with no position restraints up for a
>further 2.5 ns, and I¹m testing whether frames from the last ns produce
>similar errors i.e., does the inclusion of serine require further
>equilibration. 
>
>I would appreciate any thoughts.
>
>Many thanks
>Anthony
>
>
>
>-- 
>Gromacs Users mailing list
>
>* Please search the archive at
>http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>posting!
>
>* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
>* For (un)subscribe requests visit
>https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>send a mail to gmx-users-request at gromacs.org.



More information about the gromacs.org_gmx-users mailing list