Re: [AMBER] problem in DNA-ligand trajectory

From: Jason Swails <>
Date: Wed, 29 Oct 2014 13:22:57 -0400

> On Oct 29, 2014, at 6:54 AM, Mina Madah <> wrote:
> Dear all
> I am working on DNA-ligand system. I generate force field parameter by
> antechamber. At first I do two stage minimizations then two step
> equilibration, and finally 10 ns MD production (inputs is attached). I am
> facing a problem when visualizing trajectory with VMD, one of the
> terminal basepairs is jumped out of the box, the covalent bond which
> attached ligand to two adjacent nucleotides, is broken up, and the
> separated ligand made bad interact with other nucleotides. As a result DNA
> helix begins to deformation.
> I tried to image the ligand back into the box by ptraj, resulting
> no improvement in trajectory visualization. Extracting one snap out of the
> trajectory shows that the ligand is completely separated from DNA, and
> located in improper position out of box and helix is completely opened. But
> RMSD plot shows good shape which fluctuate around 4 angstrom.
> My question is:
> How can I perform MD for my complex to retain the covalent bonds between
> DNA and ligand?

Covalent bonds will never break in classical MD when using the Amber force field -- it is simply impossible. Either the bond exists the entire time, or it never exists, and since it is a pure harmonic potential there is no possibility of dissociation.

Furthermore, the imaging routines always image by molecule, so as long as there is a closed network of bonds that connect two atoms, those two atoms will never be imaged separately.

> Is it OK, if I use position restraint on DNA around production time ( to
> make DNA structure more stable)?

No. Based on what you’ve told me here, the problem is that there is no bond where there should be one. You can use ParmEd or cpptraj with the “printBonds” command to see if there is a bond defined between your ligand and the surrounding residues that it should be bound to.

If the bond does not exist, you should focus on fixing that problem at the model building stage (i.e., with LEaP and antechamber/parmchk).


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Wed Oct 29 2014 - 10:30:04 PDT
Custom Search