Re: [AMBER] Error while clustering on CPPTRAJ

From: <dnlfr1994.gmail.com>
Date: Tue, 19 Jan 2021 18:20:52 +0100

Thanks for the suggestions!

About the error, I think I have an explanation.
I've been doing some more quicker tests using "sieve 10" and I found the following error in the pair-distance matrix file:
#F1 F2 CpptrajPairDist
  1 11 1.2672
  1 21 1.7326
  1 31 1.8467
  1 41 2.3464
  1 51 2.6885
  1 61 3.2762
  1 71 3.2343
  1 81 3.1948
  1 91 3.0762
  1101 3.1559
  1111 2.9725
  1121 2.9481
  1131 2.8394
  1141 3.2183
  1151 2.9452
  1161 2.7912
  1171 2.2689
As you see, there is no separation between the first two columns from line 11 and below. It is the direct output of the matrixfile from cpptraj.
Now the error line says: Error: Number of columns in file changes at line 11. So this is where Cpptraj fails to read the pair-distance matrix file.

-----Mensaje original-----
De: Daniel Roe <daniel.r.roe.gmail.com>
Enviado el: viernes, 15 de enero de 2021 0:59
Para: AMBER Mailing List <amber.ambermd.org>
Asunto: Re: [AMBER] Error while clustering on CPPTRAJ

Hi,

I'm not sure I have a complete answer to your question yet, but I do have several recommendations. First, try upgrading to the latest version of cpptraj (either by using the GitHub version https://github.com/Amber-MD/cpptraj or the one from AmberTools 20).
There have been a lot of fixes and enhancements since 2016.

> autoimage
> strip :WAT
> strip :Na+

You can just use 1 strip command here, and use 'parmout' or 'outprefix' to write the corresponding stripped topology, e.g.:

strip :WAT,Na+ parmout ../top/mir_nowat.prmtop

this makes it so you don't have to use all the 'parmstrip' commands at the end.

Hope this helps,

-Dan

> rms fit :1-59
> trajout trajectories/full_prod.mdcrd
> trajout trajectories/full_prod.nc
> run
> clear all
> parm ../top/mir_solv.prmtop
> parmstrip :WAT
> parmstrip :Na+
> parmwrite out ../top/mir_nowat.prmtop
> run
> clear all
>
>
> -----Mensaje original-----
> De: Daniel Roe <mailto:daniel.r.roe.gmail.com> Enviado el: jueves, 14 de
> enero de 2021 16:52
> Para: AMBER Mailing List <mailto:amber.ambermd.org>
> Asunto: Re: [AMBER] Error while clustering on CPPTRAJ
>
> Hi,
>
> It would help if we had more information. To start, can you post your entire cpptraj input and the version of cpptraj you are using? Thanks.
>
> -Dan
>
> On Thu, Jan 14, 2021 at 8:03 AM <mailto:dnlfr1994.gmail.com> wrote:
> >
> > Hi,
> >
> >
> >
> > I’ve been doing some clustering tests on several trajectory files,
> > using the dbscan algorithm. As mentioned in the manual, I am running
> > the kdist option to choose appropriate epsilon and k values. In
> > order to speed up the process I have used the savepairdist and
> > loadpairdist commands. However, I am getting the following error
> > with some trajectories --> Error: Number of columns in file changes at line 100.
> >
> > I suspect the problem appears when loading the pairdist matrix file,
> > since the number of times this error shows up matches with the
> > number of times that my scripts loads the matrixfile.
> >
> > However, the final kdist plots (see figure below) seem to be OK.
> >
> >
> >
> > Does anyone know why this error is happening?
> >
> >
> >
> > By the way, the shape of the curves does not change much after
> > kdist=4 (as explained in the user’s guide), but the height still
> > varies remarkably even with kdist=10. I am not sure if the epsilon
> > and k values of any of these tests can be used for the final clustering or more tests would be required.
> >
> >
> >
> > Any help is really appreciated.
> >
> >
> >
> > Daniel Fernández
> >
> >
> >
> > _______________________________________________
> > AMBER mailing list
> > mailto:AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> mailto:AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
> _______________________________________________
> AMBER mailing list
> mailto:AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber

_______________________________________________
AMBER mailing list
mailto:AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jan 19 2021 - 09:30:02 PST
Custom Search