Hi Keiran,
With linear PB and idiel=1, I would not be surprised by this standard
deviation. Besides. standard deviation would be large anyhow in any sort
of MMPBSA calculations. Did you look at how the calculation converges
during one trajectory (the output energy files should be helpful for
that) ? We noticed that the st.dev. and st.error are also influenced by
the parameters used.
As for the entropy. this is yet another issue. In fact, we see that in
the "1 trajectory" scenario we are unable to get reasonable values for
the entropy for our protein-DNA systems neither using the normal mode
nor the PCA-based approach. We now have about 600 ns / system (4 x 150
ns) which is significantly longer than what we reported in the Structure
paper but still while the MMPBSA results are quite reasonable, the
entropy we get is totally unreasonable. This may lie though with the
fact that we are using a single trajectory. The entropy calculation
should in principle be very sensitive to that. But it may also point
that the entropy calculation is quite cumbersome and error-prone.
Best,
Vlad
On 09/22/2014 12:49 PM, Keiran Rowell wrote:
> Dear Vlad, Jason,
>
> Thank you for quick and helpful responses! I was just using the default linear PB solve in MMPBSA, I'll have a dig into the non-linear solvers I can see why that'd matter for a charged system.
>
> It was the large std. dev. increase which concerned me, but I'll see how that drops with tweaking. Getting okay entropic values is what I'm most concerned with, because I have FMO/MP2 calculations as well for interaction energies.
>
> Sorry my sentence should have said that I strip all solvent *and* ions before doing MMPBSA on the trajectory, though I'll run the both complexed and uncomplexed DNA with the same amount of counter-ions from now on.
>
> Thanks again,
>
> Keiran
> ________________________________________
> From: Jason Swails [jason.swails.gmail.com]
> Sent: 21 September 2014 23:42
> To: AMBER Mailing List
> Subject: Re: [AMBER] Suitability of MMPBSA to intercalator binding. 3-trajectory no improvement over 1-trajectory?
>
> On Sep 21, 2014, at 5:51 AM, Keiran Rowell <k.rowell.unsw.edu.au> wrote:
>
>> Dear Amber list,
>>
>> I'm trying to use MMPBSA to get some sort of qualitative trends in binding energies of a range of intercalators (molecules which insert between DNA bases).
>>
>> So far I've been using PB (inp=1, radiopt=0 because of new parameters) + nmode (nmode_igb=1) 1-trajectory MMPBSA, with the DNA as the receptor and my drug as the ligand. This gives an okay dG (~ -25 to -40) but little in terms of discernible trends. I thought 3-trajectory might be an improvement since there's significant structural changes to the DNA upon intercalation (not your 'lock and key' type receptor) so I thought getting separate 'receptor' values from unintercalated DNA might be more prudent.
>>
>> However upon doing this I get positive dG values and large std. dev.s, something like 20 +/- 30. I know structural values won't perfectly cancel with 3-traj, but the drop in VANDW also seems significant considering stacking interactions are what stabilise intercalated complexes. I've put an example difference output of 3-traj and 1-traj at the end of this message.
>>
>> I should note a few things which might interfere with getting good results aside from unusual 'ligand/receptor' definitions. One is my intercalators have +2 charge and so when I run the DNA on its own I have 2 extra Na+ atoms to make sure the receptor system is neutralised, but make sure to strip all solvent from the trajectories being processed. My intercalators are also dimeric so likely have a process of two insertions occurring an separate times, whereas my MD is starting from the complexed structure due to time constraints.
> Vlad gave a great response, so I’ll just make some other comments. Do not add these ions. There are some instances where a net neutral state is desirable (i.e., in some periodic simulations), but this is certainly not one of them. If you think about the thermodynamic cycle that defines MM/PBSA, you will have a tough time justifying the addition of those two ions ;).
>
> In my opinion, you should omit _all_ ions from your system (unless they are structurally relevant) and look into the non-linear PB equation to improve the treatment of the highly-charged system.
>
> HTH,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
--
Dr. Vlad Cojocaru
Computational Structural Biology Laboratory
Department of Cell and Developmental Biology
Max Planck Institute for Molecular Biomedicine
Röntgenstrasse 20, 48149 Münster, Germany
Tel: +49-251-70365-324; Fax: +49-251-70365-399
Email: vlad.cojocaru[at]mpi-muenster.mpg.de
http://www.mpi-muenster.mpg.de/research/teams/groups/rgcojocaru
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Sep 22 2014 - 04:30:03 PDT
