[gmx-users] Gromacs - Difference between target temperature and pressure in mdp and simulation

[Dallas Warren]

Re pressure, did you check the graph of what the pressure is doing?

The variation between steps of the pressure can be huge, hundreds of bar
when the average is meant to be 1. So it is not surprising that the average
isn't exactly 1. See below link for a couple of examples. Note as the
system gets larger, then the variation decreases.

https://twitter.com/dr_dbw/status/968624615063937025

Catch ya,

Dr. Dallas Warren
Drug Delivery, Disposition and Dynamics
Monash Institute of Pharmaceutical Sciences, Monash University
381 Royal Parade, Parkville VIC 3052
dallas.warren at monash.edu
---------------------------------
When the only tool you own is a hammer, every problem begins to resemble a
nail.


On Tue, 26 Nov 2019 at 06:16, Fabio Bologna <fabio.bologna3 at unibo.it> wrote:

> Dear Gromacs user-base and experts,
>
>      I need your help, as a Gromacs rookie. I've been trying to simulate
> the folding of a 20 residues mini-protein inside a dodecahedric explicit
> solvent box (30542 atoms) using Gromacs 2016.1. Because I need to start
> from the linear structure, my box is quite big ; I had to use the group
> cut-off scheme without explicit buffering instead of the default Verlet, in
> order to obtain acceptable performances. Also, I’m using the V-rescale
> thermostat (tau-t = 0.1) and Berendsen barostat (tau-p = 1.0). As in the
> title, my simulations don’t reach the intended target values of temperature
> and pressure. These are 300 K and 1 bar respectively.
>
>
>
> Because I didn’t notice earlier these issues, I let 6 production md run
> for 69 ns. The averages temperatures ranged from 302 to 303 K and the
> average pressures ranged from 0.86 to 1.1 bar. In my humble opinion, I
> don’t think it is caused by a lack of equilibration, given the small size
> of the system and the long simulation time.
>
>
>
> I checked out the previous 1ns NVT and 1ns NPT equilibration phases (each
> using the very same thermostat and barostat, when applicable) of those 6
> simulations. The average temperatures among them ranged from 300 to 302 K
> and average pressures from 0.018 bar to 1.4 bar. The temperature averages
> seems to have risen from equilibration to production MD (which completely
> puzzles me) while at least the pressure got closer to the target 1 bar
> (will it reach 1 bar later on, meaning that there are not enough points to
> make the mean value close to the target?)
>
>
>
> I ran again the NVT and NPT equilibration phases, experimenting with
> different values of tau-t and tau-p. The following results are the averages
> outputted by gmx energy (more accurate than manually averaging the values
> in the .xvg file you get from gmx energy itself, if I have understood
> correctly) at the end of the NPT equilibration.
>
>
> Regarding temperature, I got:
>
>   *   tau-t = 0.5 --> 310-315<tel:+39310315> K
>   *   tau-t = 0.1 ••• 300-302<tel:+39300302> K
>   *   tau-t = 0.01 --> basically only 300 K
>
> Shouldn’t tau-t just affect the amplitude of temperature oscillations?
> Clearly tau-t = 0.01 is the right value for my system, but as far as I know
> it is not used in productions MDs and only in equilibration phases. Is it
> still a safe value for production or does it create artifacts of any kind?
> Also, why does the most used values of 0.5 and 0.1 not give reasonable
> results? I've test it with the Verlet scheme and the simulation reach the
> correct target temperature in a total of 2 ns. Also, why did the average
> temperature rise during production from 300-302 to 302-303 K? The
> thermostat setting are the same.
>
>
>
> Regarding the pressure, I got:
>
>   *   tau-p = 1.0 ••• 0.02-2 bar
>   *   tau-p = 0.5 ••• 0.6-3 bar
>
> It seems tau-p doesn’t have any effect…
>
>
>
> Is there something I’m doing wrong or is it a normal behaviour? Especially
> regarding the pressure control, given that tau-t seems to work fine.
>
>
>
> Finally, on a different note, 1 of my 69 ns simulations crushed because of
> too many LINCS warnings. By checking the .err file, I found out that they
> were caused by a lysine and the N-terminus of the protein. In some frames
> the dihedrals of the protonated amine groups were a bit anomalous and the
> groups didn’t have a tetrahedral conformation. However the rest of the
> system was completeley fine. Also, they weren’t “misbehaving” at the same
> time. First the lysine had this issue but then the group returned to its
> correct conformation; then many ns later the N-terminus encountered the
> same problem and the total count of LINCS warning reached 1000. The crashed
> simulation was completely identical in parameters and chemical entities to
> the other 5 simulations and I didn’t see any spikes in the various energies
> of the system. Am I right to assume that it is only a numerical rounding
> error on the hardware part?
>
>
>
> I'm addition, I've attached the .mdp file I used in the production runs.
>
>
> I am sorry for this long e-mail, I hope you can help me figure out what’s
> wrong.
>
> Thank you in advance for your time and patience.
>
>
>
>
>
> <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
>
> Fabio Bologna
>
> PhD student in Nanosciences for the Environment and Medicine
>
> Department of Chemistry 'G. Ciamician'
>
> University of Bologna
>
> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>
>
>
> Scarica Outlook per Android<https://aka.ms/ghei36>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.