[gmx-users] MD runs and subsequent error

[Joern Lenz]

dear gromacs users,

i have the following problem:

i performed an md simulation of a complex of protein and DNA (not covalently 
linked) in a box of amber03_tip4p waters and the amber03 forcefield.
evereything is fine and i used the gromacs toools to extract a certain frame 
out of the trajectory file, writing it into a pdb file.

now i used pdb2gmx -merge to merge the chains in the system, since one strand 
of the DNA and the protein were recognized to be chain A, the other DNA 
strand was chain B.
I want to have the protein and the DNA in one molecule to be able to introduce 
distance restraints between certain atoms of the DNA and the protein.

running pdb2gmx and grompp aftreward run without any warnings or errors.
unfortunately the following energy minimization does not succeed because of 
potenetial energy being "nan". also an md run does not succeed, giving 
warnings about LINCS errors.

my question ow is how that can be, i mean i extracted the pdb from the trr 
file (see above) ?!?! 
Is there another way to circumvent the way of doing pdb2mgx and grompp with 
the  trr-extracted pdb file to get rid of the error (nan and LINCS) and also 
be able to say gromacs that DNA and protein are one molecule to then be able 
to introduce distance restarints in the system between DNA and protein ???
thanks a lot in advance for your advice. i hope that i soon can go on with my 
work.
greetings from warm and sunny hamburg
joern