Re: Rotate two linked proteins against each other and calculate energy profile

From: Bjoern Olausson (namdlist_at_googlemail.com)
Date: Mon Mar 28 2011 - 11:18:09 CDT

On Monday 28 March 2011 16:30:28 Ajasja LjubetiÄ wrote:
> Hi Bjoern,
>
> > I would have taken a detour if you wouldn't have mentioned ABF! Thanks a
> > lot
> > for this comment!
>
> You're welcome. Actually something similar happened to me when I was
> starting out with NAMD. After doing the basic tutorials I browsed through
> the manual, but I always skipped the colvars part, because it seem too
> complicated:)
>
For me it was the first time I had to apply a force in that way to a
simulation :)
Most of the time I gather my data via CHARMM and process them with Python :-)
But yes, I must admit that I skipped the ABF/COLVARS part because it looked a
bit complicated bin in fact it is not that hard one you get your hands on it.

> > So after playing around with NAMD CVS, GBIS, COLVARS, and ABF I could
> > verify
> > that my setup works (I just calculated the free energy profile for ethane
> > and
> > I could reproduce a nice sine like profile for -180 to 180 degree
> > rotation with a peak at around 2,7kcal for the eclipsed conformation).
> > Still I have some questions how ABF works:
> >
> > Why can't I reproduce the above when choosing "lowerBoundary 0" and
> > "upperBoundary 360" instead of -180 and 180?
>
> Have you tried setting.
> *wrapAround 180*
> **
>
It looks like this would solve the problem but:
colvars: Error: keyword "wraparound" is not supported, or not recognized in
this context.
colvars: If this error message is unclear, try recompiling with -
DCOLVARS_DEBUG.
FATAL ERROR: Error in the collective variables module: exiting.

Using NAMD CVS-2011-03-28 for Linux-x86_64-multicore-multicore-linux64-icc
Bug????

> I don't know about the height difference, it could be due to sampling? How
> does the trajectory look in VMD?
>
Looks pretty nice to me, the Methyl groups keep spinning and it _looks_ as if
they sample all conformations. I guess wrapAround would solve this abnormality
in the pmf file.

But this is not crucial to me I can live with -180 to 180 now that I know why
0 to 360 is not working.

I'll append my colvars/abf config, just in case:
colvar {
  name eprof
  lowerBoundary 0
  upperBoundary 360
  #lowerWall 0
  #upperWall 360
  wrapAround 180
  lowerWallConstant 10
  upperWallConstant 10
  outputSystemForce yes
  outputAppliedForce yes
  dihedral {
    group1 { atomnumbers {1} }
    group2 { atomnumbers {4} }
    group3 { atomnumbers {8} }
    group4 { atomnumbers {5} }
  }
}

abf {
  colvars eprof
  historyFreq 500
  hideJacobian no
  #outputFreq
  applyBias yes
}

> > If a bin is poorly sampled, one can interpret this as a highly unfavored
> > conformation or the applied force was not sufficient to push the
> > structure into this conformation and the process might not be
> > reversible. Is this assumption right (assuming that the sampling time
> > was sufficient)?
> >
> Yes, although I'm not sure about the reversibility of the process.
>
Okay, thanks.

> >
> > Is there a built-in keyword which outputs Xi as a function of time
> >
> > or can I make NAMD do so in some way.
>
> If you want the PMF as a function of time check out
> *historyFreq*
> *
> http://www.ks.uiuc.edu/Research/namd/2.8b1/ug/node56.html#SECTION0001331000
> 00000000000 *
>
Thanks. I should read the UG more carfully!

> *
> *
> *
> If you are interested in the position of the colvar as a function of time,
> then you can just plot the colvar trajectory file.
> http://www.ks.uiuc.edu/Research/namd/2.8b1/ug/node54.html#SECTION0001313000
> 00000000000
>
One option i did not miss ;-)

> Best regards,
> Ajasja
> *

Thanks again, really helpful!
Cheers,
Bjoern

-- 
Bjoern Olausson
Martin-Luther-Universität Halle-Wittenberg 
Fachbereich Biochemie/Biotechnologie
Kurt-Mothes-Str. 3
06120 Halle/Saale
Phone: +49-345-55-24942

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:01 CST