Re: ABF: Distances in colvars.traj not identical with VMD exported distances

From: Maximilian Ebert (max.ebert_at_me.com)
Date: Tue Oct 29 2013 - 10:02:05 CDT

Thanks for your reply. Just to make sure that I expressed my self correctly. In my initial simulation I used wrapall on and forceNoPBC no. After loading the trajectory to VMD I used pbc wrap -centersel "protein" -center com -compound residue -all to put all back in one unit cell and than I measured the distance. This gave huge differences in the distance. Next I used the same system wrapall on but forceNoPBC yes. This simulation pulled the substrate in the next unit cell and the substrate bonds became extremely long and the simulation crashed. Then I set wrapall off and forceNoPBC no and now when I open the trajectory with VMD and apply pbc wrap -centersel "protein" -center com -compound residue -all I get the correct distances. There is no significant differences if I run the simulation with forceNoPBC yes.

I guess turning off wrapall is the easiest and safest for ABF. Thanks for your support.

Max

On Oct 29, 2013, at 10:33 AM, Giacomo Fiorin <giacomo.fiorin_at_gmail.com> wrote:

> Hello Max, the issue is not wrapping, but the fact that in NAMD, all distances are calculated using the minimum image convention, unless otherwise specified. In VMD, most of the time the calculated distances are not PBC-aware.
>
> In your case, because I suggested you to set forceNoPBC yes, you mimicked VMD's behavior just for the sake of comparison. Evidently, you needed to disable the wrapping too because when you ignore PBCs, wrapping does affect the results.
>
> Giacomo
>
>
> On Tue, Oct 29, 2013 at 8:13 AM, Maximilian Ebert <max.ebert_at_me.com> wrote:
> I actually tried what you suggested however this didn’t solve my problem. While looking at my NAMD conf again I realized that I used wrapall on and comparing with the user guide this could cause problems in the atom group definition. I turned it off now and finally the distances measured in VMD are matching exactly the ABF output. I have a two step simulation first minimization and then using the restart files I start ABF. Together with wrapall on I think there were problems in the calculations. Could anybody explain me why wrapping in NAMD causes wrong distance calculation I thought wrapping shouldn’t have any effect other then visualization.
>
> Thank you. Max
>
>
> On Oct 28, 2013, at 5:28 PM, Giacomo Fiorin <giacomo.fiorin_at_gmail.com> wrote:
>
>> Hello Max, the distances are not necessarily calculated by VMD using the minimum image convention: can you use temporarily forceNoPBC yes in the colvars input to see if this is a PBC problem?
>>
>> Giacomo
>>
>>
>> On Mon, Oct 28, 2013 at 5:21 PM, Maximilian Ebert <max.ebert_at_me.com> wrote:
>> Dear list,
>>
>> I am running an ABF experiment using NAMD and the FF AMBER99SB-ildn. My colvars definition is added to the end of this email.
>>
>> The atom numbers are the same as in the PDB file exported from tleap while preparing the system. After running the simulation I have the impression that I picked the wrong atom numbers. I imported the trajectory to VMD and exported the distances between the two atoms (I took into account that VMD starts at zero so that the distance is actually between 7426 and 7363). The distances do not match with the distances in the colvars.traj file. Is there any possibility to verify which atoms where really picked during the simulation? Since AMBER doesn’t use a PDB file as input in NAMD there is always the possibility that the atom numbering is wrong.
>>
>> Thank you very much,
>>
>> Max
>>
>>
>> colvarsTrajFrequency 10000
>> colvarsRestartFrequency 10000
>>
>> colvar {
>> name AtomDistance
>>
>> width 0.5
>>
>> lowerboundary 3.0 ;# minimal distance
>> upperboundary 32.5 ;# maximal distance
>>
>> lowerwallconstant 10.0 ;# provides the force constant. The energy unit of the constant is kcal/mol, while the spatial unit is that of the colvar.
>> upperwallconstant 10.0 ;# same
>>
>> distance {
>> group1 {
>> atomnumbers { 7427 }
>> }
>> group2 {
>> atomnumbers { 7364 }
>> }
>> }
>> }
>>
>>
>> abf {
>> colvars AtomDistance
>> fullSamples 500 ;# Number of samples in a bin prior to application of the ABF
>> hideJacobian
>> }
>>
>>
>>
>
>

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:50 CST