Previous Thread
Next Thread
Print Thread
#13651 03/05/07 05:46 AM
Joined: Dec 2003
Posts: 94
C
cmp_fc2 Offline OP
Forum Member
OP Offline
Forum Member
C
Joined: Dec 2003
Posts: 94
Hi everyone, I have a protein on two confomers A and B, and I would like to change from A to B. I was wondering if someone knows if I only modify the phi and psi angles is possible to do such task or I have to change something else? If this is possible, is there exist any software that allow to do this? Is it possible with charmm?

Thank you for any help.

Best Regards

Joined: Sep 2003
Posts: 8,623
Likes: 24
rmv Offline
Forum Member
Offline
Forum Member
Joined: Sep 2003
Posts: 8,623
Likes: 24
Finding low energy pathways between conformational states is an active research area, and a number of CHARMM features can be used for pathway exploration, including normal mode analysis, various restrained MD approaches, etc.

I'd expect that some sidechain torsions might change as well. You should be able to use IC DIFF to compare the torsions of the two conformers.


Rick Venable
computational chemist

Joined: May 2004
Posts: 34
I
Forum Member
Offline
Forum Member
I
Joined: May 2004
Posts: 34
Hi cmp_fc2,

I have recently solved this problem and implemented the solution into CHARMM. The method is called the Harmonic Fourier Beads method and it allows to compute minimum free energy transition path ensembles between two given conformations of anything.

The method can also generate a path using a procedure, called activated evolution if the path is unknown.

The largest system that I have tested this on included 159 residues protein, NADPH and a dihydrofolate molecule. The HFB method works very well with explicit water model, and in the end provides you with the PMF or free-energy profile along the transition path ensemble.

If interested, and if permitted by THE CHARMM AUTHORITY, I can post documentation, sample input file and a patch to TREK module (c32b1) to be able to execute such an optimization.

My HFB method also works with potential energy and in that mode gives you a minimum potential energy path and its energy profile that could be compared with exact energies.

This is a major improvement to what we already published in JCP, and the new paper is currently under the final review there as well. The new implementation is called Gradient Augmented HFB method.

Best regards,
-Ilja

Joined: Dec 2003
Posts: 94
C
cmp_fc2 Offline OP
Forum Member
OP Offline
Forum Member
C
Joined: Dec 2003
Posts: 94
Hi lja.

Now the question is: What do I have to do to get your code?? :P I would like to test it on a protein of 210 aa. I have c32b1 so please tell me what to do.

Thank you so much.

Joined: May 2004
Posts: 34
I
Forum Member
Offline
Forum Member
I
Joined: May 2004
Posts: 34
Hi,

To get you started and to see if your computer resources would allow you to pull this off, I would suggest the following.

Choose the reaction coordinate space. I would recommend HARI space (see my Line-Integral paper).

Align your two conformations (RMSD best-fit mass-weighted) with that space and generate a bunch (N - something reasonable considering the extent of the conformational change, the longer the larger N should be used) of structures in between using the linear interpolation. All these are standard things you can do in CHARMM.

Then for each of the generated structures on the line run a restrained short MD about 5,000 equilibration and 10,000 steps production. You would need to use the CONS HARM the absolute atomic coordinates restraints (mass-weighted again) using the structures off the line as the references. Be clever about using the initial conditions (do not use any of the line-generated structures directly in energy evaluation, but only as a restraint reference sitting in the comparison set). For the force constant magnitude see my first HFB paper.

See if each structure's evolution takes you less than a few hours. If so you can probably pull this off. If not you probably do not need my code.

If you have a trivial transition this approach will give you the answer, if not you would need HFB to proceed. Let me know what you get.

Keep the restart files and the averaged (over the production trajectory of 10,000 steps; ~ 200 frames) coordinates for each structure.

-Ilja

Joined: May 2004
Posts: 34
I
Forum Member
Offline
Forum Member
I
Joined: May 2004
Posts: 34
Hi cmp_fc2,

I hope that you have been able to run the simulations about the linear path of your protein conformational change.

If the linear path failed to produce a smooth conformational change, consider using the gradient-augmented HFB method.

Take a look at my web-page for some additional information.
Sincerely,
-Ilja


Moderated by  lennart, rmv 

Link Copied to Clipboard
Powered by UBB.threads™ PHP Forum Software 7.7.5
(Release build 20201027)
Responsive Width:

PHP: 7.3.31-1~deb10u1 Page Time: 0.008s Queries: 26 (0.005s) Memory: 0.7553 MB (Peak: 0.8149 MB) Data Comp: Off Server Time: 2022-12-09 05:34:32 UTC
Valid HTML 5 and Valid CSS