So one weird thing I found is that when I commented out the entire [ exclusions ]
directive, the error was gone. If I kept only one line and commented out the rest, the warning came back, regardless of which line I kept. I tried to compare the output of gmx dump
for the case with and without the exclusion directives as shown below, but it is still not clear to my why this is the case:
< ld-seed = -1451233606
---
> ld-seed = -228074497
188c188
< buffer size = 8271471
---
> buffer size = 8271399
322663,322664c322663,322664
< numElements=372
< excls[0][num=12]={0, 8, 11, 12, 13, 14, 22, 27, 30, 31, 32, 33}
---
> numElements=354
> excls[0][num=11]={0, 8, 11, 12, 13, 14, 22, 27, 30, 31, 32}
322671,322674c322671,322673
< excls[5][num=14]={1, 4, 5, 6, 7, 9, 10, 16, 17, 21, 25, 26, 29,
< 33}
< excls[6][num=12]={4, 5, 6, 7, 9, 10, 16, 21, 25, 26, 28, 33}
< excls[7][num=12]={4, 5, 6, 7, 9, 10, 21, 25, 26, 28, 29, 33}
---
> excls[5][num=13]={1, 4, 5, 6, 7, 9, 10, 16, 17, 21, 25, 26, 29}
> excls[6][num=11]={4, 5, 6, 7, 9, 10, 16, 21, 25, 26, 28}
> excls[7][num=11]={4, 5, 6, 7, 9, 10, 21, 25, 26, 28, 29}
322676,322678c322675,322676
< excls[9][num=12]={4, 5, 6, 7, 9, 10, 16, 21, 26, 28, 29, 33}
< excls[10][num=14]={1, 4, 5, 6, 7, 9, 10, 16, 17, 21, 25, 28, 29,
< 33}
---
> excls[9][num=11]={4, 5, 6, 7, 9, 10, 16, 21, 26, 28, 29}
> excls[10][num=13]={1, 4, 5, 6, 7, 9, 10, 16, 17, 21, 25, 28, 29}
322695c322693
< excls[21][num=9]={5, 6, 7, 9, 10, 21, 26, 28, 33}
---
> excls[21][num=8]={5, 6, 7, 9, 10, 21, 26, 28}
322700c322698
< excls[26][num=9]={4, 5, 6, 7, 9, 21, 25, 26, 33}
---
> excls[26][num=8]={4, 5, 6, 7, 9, 21, 25, 26}
322702c322700
< excls[28][num=9]={4, 6, 7, 9, 10, 21, 28, 29, 33}
---
> excls[28][num=8]={4, 6, 7, 9, 10, 21, 28, 29}
322707c322705
< excls[33][num=10]={0, 5, 6, 7, 9, 10, 21, 26, 28, 33}
---
> excls[33][num=1]={33}
For convenience here I am attaching the figure of the ligand with atom labels.
- The two black atoms are the two atoms involved in the warning, i.e. atoms 8722 (C1 atom, which is the left one and has an index of 1 in the topology file) and 8755 (virtual site, which is the right one and has an index of 34 in the topology file).
- The three blue atoms are the parent atoms of the virtual site, i.e. indices 7, 8 and 22 in the topology file.
- The four orange atoms are the atoms involved with index 34 in the
[ pairs ]
directive, i.e., indices 6, 11, 27, and 29.
- All atoms above are in the
[ exclusions ]
directive, in which there is one additionally defined exclusion between indices 34 and 10 The atom corresponding to index 10 is in gray.
@hess do you perhaps have some insights into this issue? It is not even clear to me wether the issue lies in incorrect topology file or that GROMACS incorrectly identifies exclusions between atoms that are not even bonded neighbors. Any suggestions would be appreciated!