Re: New colvar feature request -- angleXY

From: Giacomo Fiorin (giacomo.fiorin_at_temple.edu)
Date: Mon Feb 14 2011 - 17:07:05 CST

Hi Asjasja, yes, it is possible to set a dummy atom outside the cell
boundary, but you should then use the "forceNoPBC" option to instruct NAMD
not to wrap its position to the nearest periodic image before computing
distances.

In general, I would advise against combining real and dummy atoms this way,
because everything drifts in NAMD (protein, water, lipid membranes, ...) due
to the center-of-mass motion of the system caused by the Langevin
thermostats or the external forces (see the fine print in the "COMmotion"
keyword). Even when you have a membrane protein, which should remain fixed
along z and move very slowly along x and y, the protein will actually move
very far from the center of the box where you initially set it. If you have
a soluble protein things get messy even sooner.

But in your specific case, you're already restraining the protein's position
and orientation in space as you wrote before, so what I wrote above
shouldn't be an issue.

Giacomo

---- ----
  Dr. Giacomo Fiorin
  ICMS - Institute for Computational Molecular Science - Temple University
  1900 N 12 th Street, Philadelphia, PA 19122
  giacomo.fiorin_at_temple.edu
---- ----

2011/2/14 Ajasja Ljubetič <ajasja.ljubetic_at_gmail.com>

> Dear Jérôme and Giacomo,
>
> if points A and B of the dihedral angle are the are the positions of two
> atoms and points C and D two dummy atoms, then the orientation of the plane
> normal BC will oscillate with the thermal movement of B.
> But I would much prefer to have the plane normal fixed. I could try setting
> the BC distance to a very large value (then the movement of B will not
> change the orientation of BC that much), but I don't know if it is possible
> to set a dummy atom outside of the periodic cell boundary.
>
> Otherwise I would kindly ask for a colvar component in which the two axes
> can be fixed in space (with axis and planeAxis).
>
> Thank you & best regards,
> Ajasja
>
> 2011/2/14 Ajasja Ljubetič <ajasja.ljubetic_at_gmail.com>
>
>> Thank you both!
>> This is a great idea and I think I should be able to get the desired
>> result, since I'm "cheating" a bit and restraining the rotation of the
>> protein so that I can use fixed reference axes :)
>>
>> So now I only have to derive the jacobian for the angle colvar and I'm on
>> my way.
>>
>> Best regards,
>> Ajasja
>>
>> 2011/2/14 Jérôme Hénin <jhenin_at_ifr88.cnrs-mrs.fr>
>>
>> Hi Ajasja,
>>>
>>> The coordinate you describe seems closely related to a torsional angle
>>> ("dihedral"). The torsional angle defined by points A-B-C-D is the
>>> angle between vectors BA and CD, projected onto a plane whose normal
>>> is BC. So it is likely that the existing "dihedral" component could be
>>> easily adapted to do what you need. Another consequence is that the
>>> Jacobian contribution is zero.
>>>
>>> The more annoying part seems to be the logic, since there are many
>>> ways the axes can be defined. I don't have specific usage cases in
>>> mind, so it's hard to narrow things down.
>>>
>>> Cheers,
>>> Jerome
>>>
>>> 2011/2/14 Ajasja Ljubetič <ajasja.ljubetic_at_gmail.com>:
>>> > Dear NAMD developers,
>>> > After a lot of thought I have come to the conclusion that I would
>>> really
>>> > need to sample my system along a colvar similar to the phi component of
>>> > polar coordinates.
>>> > The colvar would measure the angle between the projections of a vector
>>> and
>>> > a reference vector on a plane.
>>> > A suitable name for the colvar component would be either
>>> angleProjection or
>>> > angleXY.
>>> > It could except the following parameters:
>>> >
>>> > main and ref to define the main (measured) vector.
>>> > ref2 or axis to define the reference vector
>>> > ref3 and ref4 or planeAxis to define the normal vector of the plane.
>>> >
>>> > most probably ref3 and ref would be the same for most cases... Also
>>> perhaps
>>> > more discriptive names then ref1, ref2 etc could be used? Perhaps ref1
>>> and
>>> > ref2 for the refrence vector and planeRef1 and planeRef2 for the plane
>>> > normal?
>>> >
>>> > The main and the reference vectors are projected to the plane and the
>>> angle
>>> > between them is returned as the value of the colvar. The angle would be
>>> in
>>> > the range of [0°,360°) or (-180°, +180°] measured in the positive
>>> > orientation from the reference to the main vector.
>>> >
>>> > Would somebody be willing to help me implement this in NAMD?
>>> > I volunteer to derive the Jacobian in analytical form (needed for ABF),
>>> but
>>> > I woudl be really glad if somebody could write the C part.
>>> >
>>> > Thank you for your help & best regards,
>>> > Ajasja Ljubetič,
>>> > Young reasercher,
>>> > Laboratory of biophysics,
>>> > Institute Jožef Štefan,
>>> > Ljubljana, Slovenia
>>> >
>>>
>>
>>
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:19:49 CST