[gmx-users] trouble with a triple bond

[Alexander Bujotzek]

Dear all,

I am experiencing some trouble with a small molecule containing a C-C
triple bond, using a topology built by PRODRG beta. I intend to use
ffG43a2, with a contraint on all bond lengths.

Energy minization does not converge, as forces on the atoms involved in
the bond seem to oscillate (not very high forces, though).
Position restrained MD fails with LINCS errors,
and unrestrained MD leads to a blow-up of the system.

I guess the fault lies in the representation of the triple bond in the
topology file, as a derivate of the molecule (with a single bond in the
same place) works just fine.

Excerpt from the itp building block provided by PRODRG beta. The triple
bond is supposed to be between carbons 18 and 19:

[ bonds ]
...
  17  18   2    0.147   3422818.4    0.147   3422818.4 ;   CAQ  CAR    18 
19   2    0.122   2309822.8    0.122   2309822.8 ;   CAR  CAS    20  19 
 2    0.147   3422818.4    0.147   3422818.4 ;   CAT  CAS
...

[ angles ]
...
  16  17  18   2    109.5       520.0    109.5       520.0 ;   OAP  CAQ
CAR
  17  18  19   2    180.0  41840001.2    180.0  41840001.2 ;   CAQ  CAR
CAS
  18  19  20   2    180.0  41840001.2    180.0  41840001.2 ;   CAR  CAS
CAT
  19  20  21   2    109.5       520.0    109.5       520.0 ;   CAS  CAT
OAU
...

There is no dihedral with 18 and 19 in the middle, but I assume that is
correct as triple bonds are not supposed to be "rotatable"?
What strikes me is the high force constant on the two 180 degree angles...
may that be a source of error?

Any help is appreciated! Maybe someone even has a working template of a
correct triple bond topology at hand?

Best regards and thank you for reading

Alex

PS: I also checked ffG43a2bon.itp, but I found nothing resembling a C-C
triple bond... or maybe it's just the heat in my office.

-- 
Alexander Bujotzek
Zuse Institute Berlin
Tel. : +49 30 84185-234
eMail: bujotzek at zib.de