Hi Ming,
take a look in where your two prmtop files differ (I assume that you use
Amber9 for TI), just to check if really only the charges are different
between V0 and V1. Also, just to make sure, are your input coordinate
files absolutely identical for both processes?
You can also try to run a short simulation with ntwx=1 right were your
jobs crash to see if any atoms behave strangely.
Kind Regards,
Thomas
Dr. Thomas Steinbrecher
The Scripps Research Institute
10550 N. Torrey Pines Rd.
San Diego CA 92037, USA
On Fri, 16 Nov 2007, ming hui wrote:
> Dear Thomas,
> Yes I am removing all electrostatic interactions before removing vdW
> interactions, the problem I am facing begins right at the start in the first
> stage when only electrostatic interactions are disappearing, sorry I forgot
> to mention this.
>
> However, for the case of a TIP3 water molecule, the hydrogen atoms are
> already 'soft atoms' with non zero charges I suppose? Since the nonbonded LJ
> parameters are already zero from the start. But I have already simulated the
> system for about 10 ns before using the end coordinates as the starting one
> for my TI without any such reported errors.
>
> One difficulty I had in setting up the perturbed prmtop files for this TI
> was: in the simulation i just mentioned, the water molecule that mediated
> interactions between Tris and the protein had come from the surrounding bulk
> solvent; it was not a crystallographic water i.e. not in the starting PDB
> file. I could not use the pdb file specifying the final state of the system
> in the first simulation as the prmtop file created would not reflect that a
> rectangular box for PBC existed, or the dimensions of that box. So I had to
> start with only the protein structure, resolvate my box, then to locate the
> water molecule corresponding to the one of interest in my restart file, i
> just used >"select mol.6280" in xleap and edited the parameters of the
> selected molecule etc. I'm not sure if this would cause any problems though.
>
>
> Thank you so much for your response, I really appreciate it.
>
> Regards,
> Ming Hui
>
>
>
> On 11/16/07, Thomas Steinbrecher <steinbrt.scripps.edu> wrote:
>>
>> Hi Ming,
>>
>> I am not sure this is the source of your problems, but TI calculations
>> normally should be broken down into two steps and from what you write, Im
>> not sure if you do that.
>>
>> First remove the partial charges on all atoms that are about to
>> diappear, then in a second separate simulation remove the chargeless
>> atoms. Otherwise you will have, especially at high lambdas, nonzero
>> charges on very soft atoms, which could mean another atom gets close to
>> them and feels a huge force, resulting precisely in the vlimit error you
>> mention.
>>
>> Try out to break down your calculation into two steps by making a prmtop
>> file with all atoms still there, but zero partial charges on the ones that
>> will disappear, then use this as your end point of the first and starting
>> point of the second simulation.
>>
>> Kind Regards,
>>
>> Thomas
>>
>> Dr. Thomas Steinbrecher
>> The Scripps Research Institute
>> 10550 N. Torrey Pines Rd.
>> San Diego CA 92037, USA
>> -----------------------------------------------------------------------
>> The AMBER Mail Reflector
>> To post, send mail to amber.scripps.edu
>> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>>
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Nov 18 2007 - 06:07:37 PST
