Re: Help With colvars/abf Please

From: Jérôme Hénin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Sun Jan 30 2011 - 09:10:50 CST

Hi Angelo,

It looks like a problem with the way the colvar is defined. Most
probably, the sets of atoms for either the "ref" or "main" groups is
not what you intended. The value of the colvar at time zero (here, 46
A) should be exactly what you expect from the setup. To troubleshoot
this, check the colvars log as part of the NAMD standard output. Are
the total masses of each atom group correct? If you visualize group
"main" in VMD, remember that VMD uses 0-based indices while NAMD uses
1-based atom numbers (in VMD, they are called "serial"). You can also
visualize the reference group in VMD by loading the PDB file and
coloring the display by beta value.

Cheers,
Jerome

On 30 January 2011 14:09, Angelo Rossi <angelo.raymond.rossi_at_gmail.com> wrote:
> Hello:
>
> I am working with a lipid bilayer system with a water layer on each side of
> the bilayer.  To this lipid bilayer system,a transmembrane protein (channel
> protein) has been added.  The entire  system spans from -50 A to +50 A along
> the z-axis.  The system has been equilibrated with namd in hexagonal
> symmetry using either useConstantRatio or useConstantArea (Both of these
> work well.)  The temperature is rock solid at 300K.
>
> The idea is now to pass permeant molecules through the channel and compute
> the pmf as well as other properties.
>
> When the first permeant molecule is placed in the system above and
> equilibrated, it remains at around an average value of -20 A along the
> z-axis.
>
> Now here is the colvars input:
>
> Colvarstrajfrequency 100
> Colvarsrestartfrequency 1000
> colvar {
>   name ProjectionZ
>   width 2.0
>   lowerboundary -10.0
>   upperboundary   0.0
>   lowerwallconstant 10.0
>   upperwallconstant 10.0
> #
> #
>   distanceZ {
>
>   ref  {
>        atomsFile     file.ref
>        atomsCol      B
>        atomsColValue 1.0
>        }
>
>   main {
> #              O4     O11     O12   O13    O14    O15
>    atomNumbers 243930 243951 243972 243993 244014 244035
>        }
>        }
>             }
> #
> #
> abf {
>   colvars ProjectionZ
>   fullSamples 500
> }
>
> The above colvars input is based on the aspect of the abf tutorial which
> deals with methane hydration which is quite clever.  Thus, my reference
> coordinates are related to the top water layer around 50 A.  The idea is to
> have the permeant molecule travel through the channel and end up in the top
> water layer while starting near the bottom layer.  The oxygen atoms given
> above are at the center of the permeant.
>
> But look at the output:
>
>         # step         ProjectionZ
>            0      4.64057200542220e+01
>          100    4.62444387277113e+01
>          200    4.60430582871365e+01
>          300    4.57030585750052e+01
>          400    4.53190819052589e+01
>          500    4.50002912600680e+01
>          600    4.47016047389293e+01
>          700    4.44303855400716e+01
>          800    4.41026610823705e+01
>          900    4.38053772690827e+01
> ---------- snip snip snip ----------------------
>        49300    2.96348802222812e+00
>        49400    2.85421393614985e+00
>        49500    2.55888925656819e+00
>        49600    2.47146058653122e+00
>        49700    2.42611123875734e+00
>        49800    2.48713556793857e+00
>        49900    2.53229282527830e+00
>        50000    2.43943427282846e+00
>
> Why is it doing this?
>
> Why is the colvars calculation starting at the TOP of the channel when I
> gave it instructions to perform the calculations between -10 and 0?
>
> Of course all the other files look like the pmf file below:
> #       xi                  A(xi)
>        -10                    0
>         -8                     0
>         -6                     0
>         -4                     0
>         -2                     0
>          0                     0
>
> There are no "hits" because the trajectory is traversing the upper channel
> and not where it should be.
>
> I understand that I am doing something incorrect here and would appreciate
> any help you can provide.
>
> Kind regards,
>
> Angelo
>
> p.s.  This is one of many trial runs that I performed.
>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:23:32 CST