Re: [AMBER] velocities from and velocityautocorr

From: Ruth Helena Tichauer <rhtichau.laas.fr>
Date: Mon, 18 Mar 2019 08:36:08 +0100

Dear Daniel,

Indeed, employing the two masks results in the MSD(t) being calculated for 1 or 2 water molecules at the time.. So, as you pointed out, the issue might be the small amount of water molecules considered for the diffusion calculation.

Thank you again for your detailed reply,

Sincerely yours,

Ruth

> On 15 Mar 2019, at 20:30, Daniel Roe <daniel.r.roe.gmail.com> wrote:
>
> Hi,
>
> On Fri, Mar 15, 2019 at 10:08 AM Ruth Helena Tichauer <rhtichau.laas.fr <mailto:rhtichau.laas.fr>> wrote:
>>
>> However, I get the MSD(t) plot here attached (MSD in the xyz dimension against time). As it is not a linear plot, I don’t know how to infer from it the diffusion coefficient.. Considering an initial linear portion of the plot does not seem like an available option here..
>
> There might not be enough data going into the diffusion calculation.
> In general a lot of averaging is needed (over time and molecules) to
> get reliable translational diffusion estimates.
>
>>
>> it seems that the distances keyword, provided to the stfcdiffuion command has no effect . Indeed, while processing I get the following message:
>
> Indeed, looking at the code it seems that 'distances' is disabled if
> 'mask2' has been specified. I'm not sure why (I'm not the original
> author of the code) but it would take some re-coding to get it to
> work. Feel free to make a feature request on GitHub, but unfortunately
> the next few months are going to be very busy for me so I might not
> get to it for a while.
>
> -Dan
>
>>
>> CPPTRAJ: Trajectory Analysis. V17.00
>> ___ ___ ___ ___
>> | \/ | \/ | \/ |
>> _|_/\_|_/\_|_/\_|_
>>
>> | Date/time: 03/14/19 16:52:08
>> | Available memory: 246.294 MB
>>
>> INPUT: Reading input from 'natmm_msd-dis.in'
>> [parm ../../../native2-0.parm]
>> Reading '../../../native2-0.parm' as Amber Topology
>> Radius Set: modified Bondi radii (mbondi)
>> This Amber topology does not include atomic numbers.
>> Assigning elements from atom masses/names.
>> No SCEE section: setting Amber default (1.2)
>> No SCNB section: setting Amber default (2.0)
>> [trajin ../../../traj_nc-autoimaged/nat2-0_md1-ai.nc]
>> Reading '../../../traj_nc-autoimaged/nat2-0_md1-ai.nc' as Amber NetCDF
>> ...
>> [trajin ../../../traj_nc-autoimaged/nat2-0_md10-ai.nc]
>> Reading '../../../traj_nc-autoimaged/nat2-0_md10-ai.nc' as Amber NetCDF
>> [stfcdiffusion mask :WAT.O&!(:196-197) time 1.0 mask2 :GTP.PG lower 0.01 upper 5.0 xyz distances out natmm_msd-dis.out nwout natmm_nw-dis.out avout natmm_av-dis.out ]
>> DIFFUSION (STFC): Calculating diffusion in the xyz directions
>> Mask 1 expression: :WAT.O&!(:196-197)
>> Atoms in mask 2 (:GTP.PG) in the range 0.010 to 5.000 Angstrom will be used
>> Distances will be imaged.
>> Only the average results will be written to natmm_msd-dis.out
>> The number of atoms in the shell will be written to natmm_nw-dis.out
>> <dr^2> will be written to natmm_av-dis.out
>> The time step between frames is 1.000 ps.
>> ---------- RUN BEGIN -------------------------------------------------
>>
>> and so I don’t know how correct the obtained plot is (red line):
>>
>>
>>
>> As the version I used was the following:
>>
>> python /Users/administrateur/amber16/update_amber --version
>> Version is reported as <version>.<patches applied>
>>
>> AmberTools version 17.09
>> Amber version 16.14
>>
>> I’ve tried to upgrade AmberTools but didn’t get anything:
>>
>> python /Users/administrateur/amber16/update_amber --update-to AmberTools/18
>> Preparing to apply specific updates... please wait.
>>
>> So I've just updated to latest version that now is:
>>
>> python /Users/administrateur/amber16/update_amber --version
>> Version is reported as <version>.<patches applied>
>>
>> AmberTools version 17.10
>> Amber version 16.15
>>
>> Nevertheless, I still get the exact same results as with the previous version i.e. exact same plots for both input files and same information about imaging distances despite the distances keyword provided when analysing autoimaged trajectories.
>>
>> Thanking you and anyone for any help on any of these issues.
>>
>> Sincerely yours,
>>
>> Ruth
>>
>>> On 13 Mar 2019, at 21:56, Ruth Helena Tichauer <rhtichau.laas.fr> wrote:
>>>
>>> Dear Daniel,
>>>
>>> Thank you so much for providing me detailed insight into this matter.
>>> I’ll carefully check all your suggestions on the manual and I’ll come back to you if I have any further inquiries after trying the analysis you propose.
>>>
>>> All the best,
>>>
>>> Ruth
>>>
>>>> On 13 Mar 2019, at 18:17, Daniel Roe <daniel.r.roe.gmail.com> wrote:
>>>>
>>>> Hi,
>>>>
>>>> On Wed, Mar 13, 2019 at 12:07 PM Ruth Helena Tichauer <rhtichau.laas.fr> wrote:
>>>>> I wish to compute the velocity autocorrelation function (VACf) of water molecules within the cavity of a protein-ligand complex. I wish to use the usevelocity keyword as in http://archive.ambermd.org/201605/0236.html <http://archive.ambermd.org/201605/0236.html> but I haven’t saved any velocity information while running the calculations.. Is there a straightforward manner to re-calculate these velocities from the .nc trajectory file (using cpptraj perhaps)? FYI the trajectories I wish to analyse are 10ns long such that frames were recorded every 1 ps (I wonder if this sampling is enough).
>>>>
>>>> Unfortunately that frequency isn't nearly enough for calculating the
>>>> velocity autocorrelation function (which decays more on the order of
>>>> fs). I usually record velocities every 10 fs or so when calculationg
>>>> VAC. You're better off using the 'diffusion' command and good ole
>>>> Einstein relation.
>>>>
>>>>>
>>>>> Aside, when reading the manual concerning the velocityautocorr analysis, it is not clear to me what the maxlag is.. Neither what tstep refers to: is it the time difference for computing the autocorrelation i.e the tau in <v(t)v(t+tau)>?
>>>>
>>>> The 'maxlag' is the maximum lag you want to calculate the VAC function
>>>> for (tau in your equation). This is in general something to keep in
>>>> mind when calculation any autocorrelation function - as your lag
>>>> increases, the number of data points that go into calculating the
>>>> function at that point decreases and the function becomes "noisy". The
>>>> 'tstep' is just the timestep between frames in ps.
>>>>
>>>>>
>>>>> As the VACf is computed for a mask of atoms, and as I wish to get an estimate of the diffusion coefficient of water molecules when they are in the protein cavity (to see if this value is different from the bulk), should I select water molecules within the cavity only (I guess so)? If yes, as they exchange during the simulated time with water molecules from the bulk, should I account for the portions of the trajectory during which they are inside the cavity only (I also suppose yes so but I rather ask)?
>>>>
>>>> This is a much trickier problem - in this case I'd recommend using the
>>>> 'stfcdiffusion' command, which can calculate diffusion for atoms
>>>> within a certain distance (defined by 'upper' and 'lower') from atoms
>>>> selected by a mask ('mask2') - see the manual for more details.
>>>>
>>>> -Dan
>>>>
>>>>>
>>>>> Any help concerning any of the previous points will be much appreciated.
>>>>>
>>>>> Thanking you in advance,
>>>>>
>>>>> Sincerely,
>>>>>
>>>>> Ruth
>>>>>
>>>>> _______________________________________________
>>>>> 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
>>>>
>>>
>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org <mailto:AMBER.ambermd.org>
>> http://lists.ambermd.org/mailman/listinfo/amber <http://lists.ambermd.org/mailman/listinfo/amber>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org <mailto:AMBER.ambermd.org>
> http://lists.ambermd.org/mailman/listinfo/amber <http://lists.ambermd.org/mailman/listinfo/amber>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Mar 18 2019 - 01:00:02 PDT
Custom Search