I've got a crystal structure and there is a loop region which contains 20 aa was missed. I am planning to fill in such loop by Rosetta. However, I found that there are three different methods for loop refinement in Roseeta:
Fragments Based Loop Modeling
Kinematic Loop Modeling
Do you have any idea which one is better for my case?I've complied Rosetta by command:
scons bin mode=release extra=mpi
I don't whether those refinement can be supported by mpi multiple thread.
Thank you very much
Loop modeling does support MPI, although it's not the standard JD2 method. I'm not familiar with its details.
"loop modeling" is probably equivalent to fragment based loop modeling, by the CCD method. Kinematic is distinct.
In my experience neither work well on loops of that length. You can get an envelope of where the loop might be - what area it can sweep - but I don't usually see an energy well to a single solution. Either method works for generating such an envelope.
thank you very much for kind messages.
I am just wondering if the 'loop' actually can have some secondary structure such as alpha helix, will the kinematic protocol model such kind of conformation?
If the loop refinement protocol doesn't works well for 20aa or even 30 aa loop, is there any other alternative methods for such kind of case?
In my experience, the kinematic protocol will not generate secondary structure. The CCD protocol uses fragment insertions and thus can, sort of...but it tends to lose that structure while closing the loop.
If your loop has significant secondary structure in the middle of it, it's not a loop as far as the loop modeling protocol is concerned.
I don't know what people are using to solve longer loops like this. They may be using CCD and KIC anyway; it's just my experience that it works poorly. I guess see the most recent CASP papers?
thanks again for your kind reply. I will think about it.