I know the full amino acid sequence for my protein. I have obtained its X-ray crystal PDB structure, as well as the mtz file from the crystallisation. There are several loops and termini not resolved. Can I ask what is the best protocol to predict the missing part? I understand it would still not be fully "true" after Rosetta's modelling. I just want to have a "mostly likely" one that can be used as the starting point for my molecular dynamic simulation.
I have been told remodel is used for "design", not "predict", so not applicable to predict the missing structures.
It seems that homology modelling (e.g. RosettaCM) is recommended.
(Maybe this thread is just for a general discussion about the possible protocols available.)
Category:
Post Situation:
Remodel has lots of design features, but you can choose to not use them, say you wouldn't run it with "-find_neighbors -design_neighbors" and a blueprint with a "ALLAA" resfile code at any position. For loops you will have to use "-generic_aa G", however, for Remodel to work.
Remodel uses real fragments, so it is not totally made up. And actually, you can add posttranslational modifications that exist in vivo but the expression system did not add, so you could make a model more real than the structure (just playing devil's advocate here for fun).
Although it may pay to check for DG or DP amino acids pairs in your loops as they do funny things in vivo (isoaspartate and self-cleavage), which obviously Remodel knows nothing about.