Dear Amber users,
I am calculating the free energy change of transfering iminobiotin (IMI)
to biotin (BTN) in TIP3P water box using TI and soft core potential.
Five-nodes Gaussian quadrature is used (lambda=0.04691, 0.23076, 0.5,
0.76923, and 0.95308). This process is fine in the typical 3-step way,
i.e. neutralizing the charge of IMI, convert vdw parameters of IMI to
BTN, recharging biotin. But the free energy changes in the first and the
third steps are very large. But we only care about their difference,
which is relatively small. I think this process will bring large errors
into the final result. So I want to try a 2-step way, in which the
charges of IMI change to those of BTN and then the vdw parameters. The
prep files, frcmod files, the script for Leap module and input files for
TI simulations are attached. By comparing the prepi files, you can find
following differences:
1. there is one more hydrogen atom in IMI, i.e. H16, which is removed by
command "remove lig.1 lig.1.H16" in leap script before saving
parameter-and-topologic file and restart file for BTN.
2. some atoms with different names and types, which are changed by "set
name ..." and "set type" commands.
The first step (charge mutation) runs fine. In the second step (vdw
mutation) and lambda=0.23076, the minimization is fine. but when I try
to raise the temperature, I see "vlimit exceeded" error message and
the structure of the solute seems weird. Then I tried separated
simulations with two end-point parameters, as suggested in the mailing
list. The simulation is fine, and the structure is stable. So I have no
idea how these could happen.
Can anyone give me some suggestions to check the source of error? Thank
you in advance!
Ye MEI
- application/zip attachment: TI.zip
Received on Mon Jul 06 2009 - 09:52:23 PDT
