artificial tetramer

From: Fred
Date: 12 December 2011 19:49
Hi List,
I would like to build an artificial tetramer from a monomer PBD file. All that I have is the coordinates it self with CRYST/CELL information cards. The artificial 4-fold axis has an arbitrary orientation into the cell. I mean, its not parallel to any crystallographic axis and have to be at a certain distance of the molecule. This sounds conceptually simple, but I would like to do that in batch mode for hundreds of PDB's. Could someone, please, tell me the easiest way/program to do that?
Thanks in advance,
Fred

----------
From: Tim Gruene 

Hello Fred,

if you know the rotation matrix, you can use pdbset with its 'rotate'
keyword.
It is not clear to me whether or not you have the rotation matrix or how
you define rotation.

Cheers,
Tim

- --
- --
Dr Tim Gruene

----------
From: Fred
Hi Tim,
Thanks for your message and sorry if I wasn't clear. I don't have neither the axis orientation nor the rotation matrix. I would like to create them but don't know how and which program to use. Theoretically a have to create a axis (vector) at some distance of the molecule into the cell and give it the 4-fold propriety. Quite simple, but don't which program to use.
Regards,
Fred


Em 12-12-2011 18:23, Tim Gruene escreveu:

----------
From: Fred
at few Angstrons of the protein.

Em 12-12-2011 19:01, Jacob Keller escreveu:

How do you know where to put the axis?

JPK

On Mon, Dec 12, 2011 at 2:34 PM, Fred  wrote:

----------
From: James Stroud
This is not trivial. Assuming an arbitrary origin, the simplest 4-fold symmetry operation (4-fold rotation) has 5 free parameters (translation along the symmetry axis is irrelevant). The biggest problem is determining the values for these parameters. For example, once you apply the symmetry, your molecule may clash with its symmetry mates or not even contact them. And even if you solve this latter problem automatically (which is not trivial because of irregularity), that leaves a net of 3 parameters describing the orientation of the protomer.

James

----------
From: Fred
Hi James,
In my first post "arbitrary orientation into the cell" only means not parallel to any crystallographic axis, which would simplify things very much. I want to apply the 4-fold axis to the protein coordinates. If I have a cell and therefore an origin, I can take a point at any distance of the origin, pass a vector/axis through it and take the 3 others molecules by symmetry. That's trivial, given the point, the orientation and the property of the rotation. Don't know which program to use.
Regards,
Fred



Em 12-12-2011 19:18, James Stroud escreveu:

----------
From: Jacob Keller
Can you clarify your reason for doing this?

JPK

--

----------
From: Fred
I only want to produce an artificial tetramer.



Em 12-12-2011 19:41, Jacob Keller escreveu:

----------
From: James Stroud
Coot can do this using the Rubik's Cube principle: transform to some state where the operation can be performed, perform the operation, then transform back.

So, in coot, I would

(1) rotate the molecule to the appropriate orientation
(2) move to the appropriate place in the unit cell
(3) change the symmetry to P4
(3) apply P4 symmetry
(4) change the symmetry back to whatever
(5) move the tetramer back to the appropriate place in the unit cell
(6) rotate the molecule back

Then, your original molecule will be in the original place surrounded by it's symmetry mates.

James

----------
From: Eleanor Dodson
I am a bit puzzled - do you know the direction of the rotation axis which can be converted to polar phi/psi angles, a mid point you want to rotate about, and the distance of the centre of mass of your molecule from that point, from which you could find the tranlation shift?

If so I could design a script for pdbset
rotat polar phi psi 90
tran tx1 ty1 tz1


Then
rotat polar phi psi 180
tran tx2 ty2 tz2


And
rotat polar phi psi -90
tran tx3 ty3 tz3


You would have to reassemble the tetramer from the 4 copies.
Eleanor

----------
From: Eleanor Dodson
Do you mean just any tetramer?
If so, Why not use an existing one - eg haemoglobin..
Eleanor

----------
From: Fred
Dear CCP4bb list,
Thank you very much all of you who have answered my post. I'm really sorry if I was unclear. Such operation is so unusual that I could be able to express myself appropriately. From quick reading some replies (James Stroud and Guillaume Ponchel), it seems is possible do build artificial tetramers with Coot. Several problems have been raised like clashes, unusual interfaces and so on.  A second step would be to take Coot's rotation and translation matrix and apply it to all pdb's in batch mode with pdbset. All pdb's are superposed by a common sequence region, which also will be part of the tetramer interface. I'll try to make things work.
Once more, sorry for any inconvenience and thank you very  much.
Kind regards,
Fred

----------
From: Tim Gruene
Hi Fred,

this sentence of yours, "All pdb's are superposed by a common sequence
region, which also will be part of the tetramer interface." probably
hides the information which would be necessary for a reasonable answer
to your question.

If you still are stuck, you might post again with a more detailed
description of what you mean.

Cheers, Tim

----------
From: Fred
Hi Tim,
Thanks for your replay. All pdb monomers have the same primary sequence and a perfect matching long helix, which I have used to superpose the coordinates. Such helix is almost straight so that, the idea would be to create a vector along the helix main axis, shift this axis to a some distance (perhaps minimizing clashes) and apply 4-fold rotation. A second step would be to take these into pdbset to make things in batch mode. It sounds simple, but don't know the easiest way/programs to do that. I can do just the basics in Coot. I remember that Xfit had some options to trace vectors inside a cell and give it rotation properties. However, Xfit seems to be frozen and integration with pdbset would be painful.
Regards,
Fred

Em 14-12-2011 07:32, Tim Gruene escreveu:

----------
From: Eleanor Dodson
This sounds as though you will create a 4-helical rod?

There are lots of 4-helical bundles but they arenot completely rod like - there is a twist in most I have seen.
Maybe you should do secondary structure matching to a suitable 4-helicl bundle, match your helix in turn to each of the bundle helices.
Eleanor

----------
From: <Herman.Schreuder
Dear Fred,

In this case, I would take one monomer and edit the CRYST1 card of the
pdb to have a sufficiently large tetragonal unit cell.
Next I would manually align the helix of one monomer with the fourfold
axis, e.g. the Z-axis. If you display the unit cell, you will get e.g.
the 0,Z,0 axis to guide you.
Then I would manually translate the monomer such that it forms a nice
tetramer with its symmetry mates and would write out these symmetry
mates.
Once you have one good tetramer, you can (or have somebody) write a
script to superimpose all your other monomers onto this template.

My two cents,...
Herman

----------
From: Tim Gruene
Hello Fred,

as far as I understand your problem I am not aware of existing programs
to carry out the tasks. To automatise the process I would write a
program myself which takes the PDB-file and the reside numbers which
make that helix as input and:
1) determine the axis of that straight helix (probably by determining
the moments of intertia, see e.g. Gruene, T. & Sheldrick, G. M. (2011).
Acta Cryst. A67, 1-8, section 2.2.1 with x_o the centre of mass.
2) copy the monomer by 3 rotations of 90 degrees about this axis to get
the tetramer.
3) move each of the four monomers stepwise perpendicularly to the
rotation axis until no more clashes are there (mask the volume of one
monomer and check each atom/atom volume whether it falls into that
volume. Because of the 4-fold rotation axis you only need to mask one
monomer and check against its next and its opposite neighbour).

The tricky bit w.r.t automatisation would be to determine the helix
axis, unless you already have a list of composing residues for each
PDB-file.

That's probably nothing I would finish in a week's time, though...

Tim

----------
From: Eleanor Dodson
An old old trick for getting a straight helix along a crystal axis using the GUI.
Run the Amore TABLING step - it is part of the GUI with the helix as your "model". (See molecular replacement - amore suite.)

Amore calculates the ppl axes and will align it along the Z axis ..

Thank you Jorge..
Eleanor

the moments of intertia, see e.g. Gruene, T.&  Sheldrick, G. M. (2011).

----------
From: Phil Evans
I've done this sort of thing (but different) using the Clipper libraries in C++. I found it easier to write short programs to do specific tasks rather than try to write generic all-purpose tools
Phil







From: Fred


Hi there,
Thanks everyone who answered the post entitled "artificial tetramer" . This is a resume of things I have done to build artificial tetramers. The following are the steps I have used:
1) Using your helix as model, run Amore scenario "get origin shifted model" with option "Reorient Trail Model" checked  (From Eleonor Dodson):  this will give your helix perfectly oriented into a P1 cell;
2) With pdbset, translate your helix by the desired distance along X and apply the desired symmetry to Z (P4 in my case): this will give your template;
3) Rename the chain´s template and write a simple script combining "lskab" and "pdb_merge" in a FOR loop to superimpose to and merge the mobile pdb with your template one chain after another.
In principle I was thinking about get a arbitrary point into the cell, pass a vector through it and apply the desired symmetry. However, unless XFIT, I don't know which program else has these capabilities. So the steps described here in are not of general application, but has solved my problem.
Best wishes and Marry Xmas,
Fred