[gmx-users] Free energy of discharging and then recharging not zero

[David Mobley]

Bob,

> Hello everyone,
>  I'm trying to calculate the free energy of binding of DNA bases on a
>  carbon nanotube. I'm running some tests to make sure that I'm doing
>  everything correctly. One thing I tried was turning off all the atom
>  charges of the DNA base and then turning them back on again.
>  Theoretically, the free energy changes of these two processes should
>  be equal and opposite and thus sum to zero. However, this is not what
>  I'm finding.
>
>  For guanine, I get a free energy change of 648 kJ/mol and -618 kJ/mol
>  for turning off and turning on the charges, respectively. Obviously,
>  they are not equal by 30 kJ/mol, which seems pretty big. I have done
>  some error estimation using the g_analyze -ee program. One thing I
>  find strange is that the error estimates in dV/dl for TURNING ON the
>  charges is large (over 2) and do not even converge for a 7.5 ns
>  simulation. In contrast, the error in dV/dl for TURNING OFF the
>  charges converges extremely quickly (using small block sizes of 50 or
>  less) and is smaller at 0.3. So it seems like I have some sampling
>  problems with the TURNING ON portion. Is there some reason why you
>  must sample a longer trajectory when turning on the charges?

Er, can you provide a few more details here? i.e., when you say "a 7.5
ns simulation", do you mean 7.5 ns at each lambda value? (I appreciate
that you're doing the error analysis separately at each lambda,
though).

Are you restraining the base to be near the nanotube in a particular
orientation? For example, I could imagine that if the base had drifted
away from the nanotube after turning off the charges, then it might be
hard to converge the calculation for turning on the charges (and the
calculation for turning off the charges would probably not be
converged either, though it might appear converged).

The only reason I can see why you might have different convergence
properties for turning on the charges than turning off the charges is
if the starting state in the two sets of simulations is different (or,
if you are starting the simulation at each lambda value from the end
point from the preceding lambda value). Are you doing one of those
things?

David

>  I'm following the procedures of
>  http://www.dillgroup.ucsf.edu/group/wiki/index.php/Free_Energy:_Tutorial
>
>  Does anyone know the reason for the discrepancy between these two
>  (seemingly identical) processes?
>
>  Thanks,
>  Bob
>  _______________________________________________
>  gmx-users mailing list    gmx-users at gromacs.org
>  http://www.gromacs.org/mailman/listinfo/gmx-users
>  Please search the archive at http://www.gromacs.org/search before posting!
>  Please don't post (un)subscribe requests to the list. Use the
>  www interface or send it to gmx-users-request at gromacs.org.
>  Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>