You are here

Model two domain protein

2 posts / 0 new
Last post
Model two domain protein


I want to model a protein that has two domains. I have two separate domains modeled using Rosetta. I want to merge these two domains based on the cst file that gives inter domain distance constraints.

What are some of the best ways to model it? These domains are  asymmetric. Can I use RosettaCM to do the modeling? Or should it be more like fold and dock?


Post Situation: 
Sat, 2017-05-06 13:21

Fold and dock probably isn't what you want. Fold and dock is specific for ab initio folding of symmetric proteins. If you have two asymetric monomers (or two domains of a single protein), it probably isn't going to work out.

If you have templates for each of the two domains, then yes, you probably want to use RosettaCM, rather than ab initio. If you have decent templates, you'll get much better results with homology modeling.  Ab inito is something of a last resort if you don't have any templates, or the templates aren't actually close to the actual structure.

RosettaCM should be able to handle having both domains in the same run - you just need to make sure your alignments put the correct template on the correct domain. This will work best if you have at least one template that shows how the two domains are oriented with respect to each other, to help RosettaCM place each in the combined run. It's a little trickier if there isn't that template information. Inter-domain constraints will certainly help, and it might be worth a try, but don't be too surprised if things don't work out too well.

Another approach (and one that's more likely to give decent results), is to treat the two domains as two independent chains, and then do protein-protein docking of the two. (With the constraints.) This should find the most likely inter-domain orientations of the two domains. You would just need to filter the results with an eye toward the fact that the two domains will need to be covalently connected later. (e.g. make sure the termini distance isn't too much, and that it's not trying to bury a terminus in an inaccessible location, etc.)

To actually do the connection of the two domains, I would go back to RosettaCM. Now you have a template (your docked model) which show the inter-domain orientation. RosettaCM should be able to use this to build in the missing loop region connecting the two domains. -- You'll probably want to do RosettaCM several times, each with different possible domain orientations, then pool the resulting structures and pick the most likely results.

Mon, 2017-05-15 08:54