VMD-L Mailing List

From: Pawel Kedzierski (pawel.kedzierski_at_pwr.wroc.pl)
Date: Fri Oct 21 2011 - 04:26:51 CDT

Dear Bruno,
forgive me if I what I write below is obvious to you but I am not quite
certain about that...

W dniu 20.10.2011 13:15, Bruno Luís Pinto de Oliveira pisze:
> Dear Pawel,
>
> Thank you very much for your reply.
>
> Maybe I'm not being clear enough and so I am not getting any help.
>
> I published in my webpage the thermodynamic cycle of my transformation to be more easy to understand.
>
> http://www.boliveira.com/NAMD.html
>
> As I said in my transformation I did the set up such that the Re(CO)3 core vanished.
>
> I was guessing that during the simulation the amines that are coordentated to the Rhenium will turn flexible and interact with the protein. What I see during the simulation is that the amines are still coordenated to the Rhenium and the rhenium core did not disappear.
The setup you describe is what you did in NAMD, but you visualize the
results with VMD which is not aware of that setup.
What you read into VMD is the topology (PSF file) and coordinates (for
example, DCD file). If you read in just a PDB file, the topology is
guessed so you *should* read in both PSF and DCD to visualize the
trajectory.
Neither if these files contain any information of your FEP setup. In
fact, due to technical limitations both of the forcefields used in NAMD,
and of the VMD program, the topology of the system (number and types of
atoms,bonds etc.) *must be constant* over the entire trajectory - both
during simulation and visualization.

What NAMD does during FEP simulations is that the both vanishing and
appearing parts of the system are present and included in all
calculations, but only the energy and force terms related to these parts
are multiplied by your lambda parameter or (1-lambda), respectively.

VMD has no knowledge of all this and it faithfully displays all atoms
over the entire trajectory.
> Now I doubt if I did correctly the set up or if the set up is OK and the Re(CO)3 is presented during the simulation but in fact is not interacting with the protein ???
>
> Lucky (or not) the caculated results are in good agreement with the experimental ones.
>
> I found this information in the NAMD "manual":
>
> "It is also worth noting that the free energy calculation does not alter intramolecular potentials, e.g.bond stretch,
> valence angle deformation and torsions, in the course of the simulation..."
>
> http://www.ks.uiuc.edu/Research/namd/2.6/ug/node36.html
>
> That´s why I didn´t see the arms that are coordenated to the metal turning flexible?
Even with the core displayed all the way by VMD, the interactions with
the vanishing part should get softer and softer so at the end of your
simulation (where lambda=0) you should see that these atoms are not
"felt" by the "arms" and you probably should see the "arms" (more)
flexible and their or other atoms probably overlapping with the core.
If not, it might indeed mean that something is wrong with your results.
Note that FEP modeling is inherently intricate and the results are
sensitive to the setup. As far as I remember, good equilibration and
sampling (in terms of lambda steps) of the beginning and the end of the
transformation is especially important.

Having said that, the best what you can do with VMD visualization is to
/mimic/ the transformation occuring over the trajectory.
This is what I proposed in my previous email. Because VMD has no
knowledge of the lambda parameter, you should write a Tcl function which
will calculate the lambda from the trajectory frame number (for example;
or you can provide a file mapping frame to lambda), and use this lambda
to make the graphical representation of the core to vanish; for example,
by scaling the atom and bond radii by lambda, or by making the
representation gradually more transparent.
HTH,
Pawel Kedzierski

P.S. My last email bounced off your mailbox due to quota exceeded.
> Can anyone help me with some comments ??
>
> Thank you very much for your help
>
> Bruno