[AMBER] Fwd: Re: RE: question on section 2 of MM-PBSA tutorial

From: Andrew Voronkov <drugdesign.yandex.ru>
Date: Sat, 17 Apr 2010 00:57:54 +0400

Thank you for the answers. I now understand better how it was done.

"> > Maybe I can use production run for some of these data, but the
> > conditions won't be the same as in the section 1.
> I am not sure exactly what you mean here."

I acutally didn't understood what was the purpose for density run as far as it was just restrained molecular dynamics. While in some tutorials :
the constant volume molecular dynamics was performed at this stage with fixed biopolymer by weak restraints:
which is corresponding to heat run here:
how different is restraintmask with force constants from such approach of fixed biopolymer with weak restraints as in polyAT_wat_md1.in?

Btw also I a little bit don't understand the purpose of NMR restraints usage in the heat.in? There are no references to NMR in the text.

Aslo in MM/PBSA tutorial no minimization without restraints of the whole system was performed as here:

So at least I plan to add minimization without restraints to the computations in MM-PBSA tutorial, before heat.

PS I am just try to fit MM PBSA tutorial and DNA tutorial parameters to each other.

Best regards,

08.04.10, 22:42, "Ross Walker" :

> Hi Andrew,
> The question of 'equilibration' is a very subjective one that we could no
> doubt debate for hours. Are any of our simulations ever truly equilibrated?
> Who knows... Probably not, would be my answer.
> With aside though, referring to your question.
> > it is written in section 2:
> > "It is essential, for good results, that our system still be exploring
> > equilibrium phase space during the production phase. We will check this
> > in the same fashion as we did for the last equilibration step by
> > plotting the density, temperature, total energy and backbone RMSD."
> > What does this phrase exactly means?
> What I am referring to here is that your results should NOT be a function of
> your starting conditions. In other words if you find that your results
> change markedly by starting with a slightly perturbed starting structure
> then probably you have not equilibrated / run for long enough.
> > Have you made here also 2 ns simulations for heating and density
> > equilibration with weak restraints as in section 1? Or you've just made
> > plots from production runs?
> The plots are just for the production runs. Essentially:
> process_mdout.pl prod1.out prod2.out prod3.out prod4.out
> xmgr summary.DENSITY
> Note the x axis does not start from zero since 600ps of calculation (heating
> and density equil) had already been done prior to what is being plotted
> here.
> The same is true for the RMSD plot, although the key is that the structure
> being fit to is still the original structure.
> > So that was done above 4 production runs?
> In the case of the tutorial, nothing. We just took the 2ns of production
> simulation done and use that for the rest of the work. If you are doing this
> on a real world system you should probably look carefully at the literature
> to see what others have done before proceeding further.
> > I am not sure how much time this will take, but if I have production
> > run of 20 nanoseconds these heating, density, constant pressure
> > equilibrations can take quite a long time. Do I need to run them again
> > or that was mentioned just about plotting of data from the production
> > run? Maybe I can make them for some last parts of the system, like
> > last 2-5 nanoseconds?
> Possibly... The plots are easy - just use the process_mdout.pl script. The
> time consuming part is the mmpbsa calculation. In an ideal world you would
> probably look at say 10 to 15ns and 15 to 20ns. You should find, if all your
> calculations are converged and run for long enough that you get identical
> answers. Of course in reality you won't so it becomes a subjective call.
> There is always a tradeoff between how old one wants to get waiting for the
> calculations to run and how desperately one needs to publish the work.
> Unfortunately one just has to make a judgement call on this. Just be sure
> you can back it up with data showing the results look to be converged, or
> are converging. E.g. if you plot the binding energies you get as a function
> of the number of snapshots included, from say 1 to 10000 does the function
> look to be converging. How does what you have done compare with the work and
> experiences of others in the field, etc etc.
> > Maybe I can use production run for some of these data, but the
> > conditions won't be the same as in the section 1.
> I am not sure exactly what you mean here.
> Note, the plots shown in this tutorial aren't of any use other than just
> visually eyeballing that things look okay. I.e. the RMSD hasn't gone haywire
> because the system unfolded etc etc.
> Good luck,
> Ross
> /\
> \/
> |\oss Walker
> | Assistant Research Professor |
> | San Diego Supercomputer Center |
> | Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
> | http://www.rosswalker.co.uk | http://www.wmd-lab.org/ |
> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
> be read every day, and should not be used for urgent or sensitive issues.

-------- Завершение пересылаемого сообщения --------

AMBER mailing list
Received on Fri Apr 16 2010 - 14:00:02 PDT
Custom Search