[gmx-users] Re: Diagnosing + system blowing up

Mark Abraham Mark.Abraham at anu.edu.au
Wed Aug 1 00:19:13 CEST 2012

Shima Arasteh wrote
> Thanks for dear Mark's suggestions.
> What's the typical solution to fix such errors of grompp?

As I've said lots of times, you need to go and look at the atom types you're
using, what interactions are defined for them in ffbonded.itp and perhaps
reassess those types. Perhaps you can invent some interactions by analogy.
Perhaps you can parameterise some. Perhaps you should give up and use

Shima Arasteh wrote
> I don't have any idea to do what, so erased the lines defined in output of
> grompp,

As Justin said, blindly deleting things is a terrible approach. This assumes
that pdb2gmx created interactions that weren't useful to your model physics,
given the fact that you've forced it to use atom types that the makers
didn't intend. It's much more likely you're doing something wrong than
pdb2gmx is doing something wrong.

Shima Arasteh wrote
>  then I went through the NVT equilibrium, it didn't stop by multiple
> interaction errors. I don't know what will happen for the rest of the
> investigation, put the protein in bilayer and then Umbrella Sampling and
> .............  .

No point worrying about that yet. You're trying to run a simulation on a
species that your force field didn't anticipate. So you've got to fix that,
or choose a different force field, or a different simulation.

Shima Arasteh wrote
>  ERROR 1 [file topol.top, line 414]:
>       No default Bond types
> And wrote line 414 is :
> line 414 of topol.tp:
>    [bonds]
>          1     2     1

Think about this. grompp couldn't find a bonded interaction of type 1 that
matched atoms 1 and 2. That's your formyl carbon and hydrogen. That's not
suprising - you're using the C atom type, which you will likely see in
atomtypes.atp described as "carbonyl C, peptide backbone". Only if someone
really wanted to model a formyl peptide cap would there be a bonded
interaction defined from that C type with a hydrogen atom type. Anyway, if
you can find a bonded interaction type between H and C atom types that seems
to you sufficiently analogous, then you might copy that in ffbonded.itp for
the atom types in your molecule. Likewise for angles and dihedrals. Probably
you should give up and use acetyl.


View this message in context: http://gromacs.5086.n6.nabble.com/Diagnoding-system-blowing-up-tp4999804p4999901.html
Sent from the GROMACS Users Forum mailing list archive at Nabble.com.

More information about the gromacs.org_gmx-users mailing list