You are here

modeling an extracellular loop of a transmembrane protein

6 posts / 0 new
Last post
modeling an extracellular loop of a transmembrane protein

Dear all,

I built a model for a membrane protein and know I try to refine a long extracellular loop using NG-KIC modeling. I noticed that some sgments can reach the outer transmembrane space. I am trying now to use constraints to penalize this behaviour. In your opinion what is the best type of constraints I can use to achieve this.


Post Situation: 
Fri, 2016-08-12 06:16

There is another long thread somewhere around here on using constraints to preclude loops entering the membrane.  I can't find it.


The short answer is, you can manually set up a series of repulsive coordinate-based constraints to forbid your loop wandering towards where the membrane is.  Figure out the membrane plane in your protein's coordinate frame and write a script to write constraints forbidding the loop CAs from being in that region. 

The newer membrane scorefunction defines a virtual membrane and may be useful for this as well.  I have contacted those authors for ideas.

Sat, 2016-08-13 11:06

I would suggest to first visualize the membrane to make sure your protein is embedded the way you think it is. The next question is whether you are using the membrane scorefunction while loop building. If both of these things are correct, then Steven's advice of using atom pair constaints or coordinate constraints might work best. 

Mon, 2016-08-15 20:01

By "make sure your protein is embedded ..." do you mean that the priciple axis of the protein should be aligned to the Z axis ?

To activate the membrane score funtion should I add this option ?


-score:weights ./membrane_highres_t1.wts



To make the constraints, I think about using something like the following in the constraints:


Dihedral CA 57 CA 507 CA 797 CA 66  PERIODICBOUNDED 0.1 0 3.14 2.0 0.5 tag

here residues 57, 507 and 797 define the parallel plane to the membrane bilayer which are not part of the loop segment, and so the dihedral angle formed with the residue 66 which do part of the loop) will be allowed to move between 0 and pi.

Should this work for my case?

Best regards


Tue, 2016-08-16 08:46

I don't know if the dihedral constraint allows improper torsions like that...but if it does that's a brilliant, simple solution!

I assume Julia will reply on the scorefunction query.

Tue, 2016-08-16 10:05

Rosetta doesn't use an absolute membrane orientation. That is, the membrane normal isn't confined to the z-axis.

We currently have two differnent membrane formalisms in Rosetta. There's the old membrane code, and the new membrane framework.

The old membrane code should spit out PDB-like descriptions of the membrane oritentation to the tracer. I belived you would just need to grab the version which is printed just prior to the structure output, and then load that into something like PyMol to show how the membrane is oriented.

As understand it, the new membrane framework should represent the membrane orientation as a "virtual residue". If that isn't being output in your structure, you may be able to include it with the "-output_virtual" commandline option.

Mon, 2016-09-05 10:59