Dear researchers
I am facing this error during the free energy calculation of protein-ligand for which I have applied restrained to some atoms of protein residue and ligand.
A list of missing interactions:
Not all bonded interactions have been properly assigned to the domain decomposition cells
A list of missing interactions:
Angle Rest. of 2 missing 1
Dih. Rest. of 3 missing 1
Molecule type 'Protein'
the first 10 missing interactions, except for exclusions:
Angle Rest. atoms 3437 7908 7905 7908 global 3437 7908 7905 7908
Dih. Rest. atoms 7905 7908 3437 3439 global 7905 7908 3437 3439
-------------------------------------------------------
Program: gmx mdrun, version 2018.1
Source file: src/gromacs/domdec/domdec_topology.cpp (line 436)
MPI rank: 0 (out of 5)
Fatal error:
2 of the 64584 bonded interactions could not be calculated because some atoms
involved moved further apart than the multi-body cut-off distance (1.223 nm)
or the two-body cut-off distance (1.223 nm), see option -rdd, for pairs and
tabulated bonds also see option -ddcheck
Could anybody please suggest how should I fix this error? I would be really thankful.
Regards
Thanks.
Sadaf
Hi,
Domain decomposition (DD) is really annoying. In your case, you could try -rdd 1.4, which limits the minimum cell box size for DD higher than the cut-off.
Also, you can read gromacs manual (p. 246) for more options.
Best,
The decomposition indeed is sometimes more warning/error-claiming than other software. However, often this error/warning comes from the problems in your simulated system. Therefore, you should take a look of the physical properties of your system and identify the problem, rather than just perform simulations as long as it can run. Otherwise, maybe after several weeks waiting, the outcome of your simulation is wrong, which is obviously not satisfactory.
Thank you Antoniel Augusto Severo Gomes and Zhaoxi Sun. I tried to specify -nt along with -pin on and my simulation completes without any warning of missing interactions. I am not exactly sure that is this the correct way. I have used the same system for a simple MD simulation and didn't receive any such error. This problem I am facing in free energy calculation.
I need your suggestions for my improvement.
Many thanks.
Sadaf.
Sadaf Rani,
Yes, you can use -nt. In short, DD divides the box in N small boxes, with N equals to number of MPI ranks. If you decrease the number of ranks, it will take more time to finish your simulation (however, you prevent such errors).
I just got a case like that with systems really similar, and I had to use -rdd 1.4 and 1.6 for each one.
If you want to improve performance, try to play with -rdd or -dds.
Best,
Antoniel.
- Badges
- Science topic