Modeling circular permutation and domain insertion

I’m not aware of a good way to do this in 3.2. I’m aware of a few functional, but not stellar, ways to do this that will release in 3.3.

You could consider doing Relax with fairly tight dihedral constraints on the protein cores, and no constraints on the new linker region?

What you really want to do is loop modeling, where the domain insert is treated as a big fixed portion of loop (no internal degrees of freedom changed). This is not possible without modifying the loop modeling code. If you go in and change how the MoveMaps are set up to fix the conformation your inserted domain, you can do CCD loop modeling (and with some more work, KIC too). I can point out the functions to work with if you want to try this, but I can’t write the code for you.

Failing that, you can try Relax mode. You can write a constraint file (http://www.rosettacommons.org/manuals/archive/rosetta3.2.1_user_guide/constraint_file.html) that includes a DihedralConstraint for each torsion in the core of your protein, and in the inserted domain, but no constraints on the “loop” residues. Relax mode will try to model all residues, which is inefficent, but the constraint will prevent it from making changes outside the loop region. This is likely to be very slow compared to code written for the purpose, but it’s the only thing I can think of that you can do in 3.2.

There may be a way to do it with the abinitio “broker” code in 3.2 but there is no documentation for it.

The length of the loops won’t matter for the relax I’ve described above. Normal loop closure does not work well on loops that long. Nothing in Rosetta (nothing I’m aware of in other programs either) works well on unstructured regions of that length; for structured regions ab initio prediction is your best bet. Do you expect your loops to be internally structured or not?

“loops to be internally structured ??”

Is it all loop, or does it have stretches of helix and sheet? Is it fairly rigid in the folded protein, or is it completely flexible? If it’s flexible, then you can sample where it may flex to, but there’s no point in trying to find one single structure because it won’t have only one anyway.

With loops that long, you have another option besides relax; you may be able to do docking on your two rigid domains, then do loop modeling between fixed endpoints once you have the domains in place. (You’ll still need fixbb separately to do design). What you really need is a custom program; you could look into writing yourself exactly what you want with RosettaScripts or PyRosetta.

Modeling loops should proceed okay with the normal loop modeling executeable. Its function is insensitive to loop length (although it isn’t going to work well on loops that long, as I’ve already described.)

I’m not proficient at either RosettaScripts or PyRosetta.

Just to clarify, PyRosetta and Rosetta are basically the same thing – PyRosetta is simply a version of the Rosetta library wrapped to be available to Python programs, as opposed to the native C++. RosettaScripts is an application in the C++ version of Rosetta which allows you to build custom protocols via XML files.

RosettaScripts works well if you can conceptualize your proceedure as a sequential application of the basic mover building blocks which are currently available in RosettaScripts (see http://www.rosettacommons.org/manuals/archive/rosetta3.2_user_guide/RosettaScripts.html for a list of available components in 3.2.1)

If you need something that’s not currently available in RosettaScripts, or your proceedure is more complicated than a simple sequential application (e.g. complex conditionals, complicated looping structures), then PyRosetta would be a better choice. (PyRosetta can handle the simple sequential protocols as well). You have access to almost all of the Rosetta library, but through a convenient, full-featured scripting language (Python).

A third option, if you’re especially hard-core or doing something that’s not possible with RosettaScripts or PyRosetta, is to write a new Rosetta executable in C++ (or modify an existing one) and compile it, linking in the Rosetta library. This gives you the most flexibility with Rosetta, but it’s also the most prone to frustration. I’d definitely recommend trying RosettaScripts/PyRosetta first.