Re: [AMBER] Exiting due to the presence of inconsistent SHAKEH hydrogen clusters. with pmemd.cuda

From: David A Case <>
Date: Tue, 22 May 2018 15:21:31 -0400

On Tue, May 15, 2018, David Sez wrote:
> - the psf and pdb files created with psfgen, solvate and autoionize plugin
> of vmd.
> - the inpcrd and prmtop generated with parmed/chamber of Amber2017
> - a min.rst and files to run equilibration.
> - a folder named topo with the topology and parameter files
> When using ntc=2, ntf=2, I run "pmemd.cuda -O -i -o mdout -c min.rst
> -p e175l_.prmtop" and the following error occurs:
> 31: 60 - 62 and 34: 62 - 63
> Hydrogen atom 62 appears to have multiple bonds to atoms 60 and 63 which is
> illegal for SHAKEH.


With Jason Swail's help, I think I've tracked down the problem.

According to the cA.psf file you sent, several files were loaded:

 REMARKS original generated structure x-plor psf file
 REMARKS 2 patches were applied to the molecule.
 REMARKS topology topo/top_all27_na.rtf
 REMARKS topology topo/top_all27_prot.rtf
 REMARKS topology topo/toppar_all27_na_nad_ppi.str

In particular, the top_all27_na.rtf file has incorrect entries for atom
types NR1 and NR2:

MASS 106 NR1 14.007000 H ! For nitrogen in imidazol (NF)
MASS 107 NR2 14.007000 H ! For nitrogen in imidazol (NF)

This makes the NR1 and NR2 atoms appear to be hydrogens, although the
masses are those of nitrogen.

(The same error appears in top_all27_prot_na.rtf).

In your protein, your HSE and HSD (histidine) residues have these atom
types, but the prmtop file that parmed/chamber creates gives them an
atomic number of 1, and I'm suspecting that comes from the two errors
listed above.

I'm guessing that you may have used top_all27_prot_na.rtf as the RTF
file you gave to chamber. If instead, I use top_all27_prot.rtf, and
rebuild the Amber prmtop file, pmemd.cuda seems happy.

So: can you try the same thing, and/or fix the problem shown above? Let
us know if you still have problems. Also, you might want to let Alex
MacKerell or other CHARMM people know about this problem.

Thanks for sending the detailed report. It took some digging to track
this down, since the protein RTF file is OK, and your system is a


AMBER mailing list
Received on Tue May 22 2018 - 12:30:01 PDT
Custom Search