Currently I was solving a low resolution protein crystal, and I can build a major part of my protein. And then,I was using rosseta to generate a 9ers, and want to choose a proper fragment as a model to build the poorly identified part of my protein manually. So, I have to pick the 200 fragments one by one, by get the pdb code, and download the pdb files and finally find the corresponding sequence to see if the shape of model fits my protein and the poor density. And this process is very time-consuming.
I was wondering whether there is a software to view the fragment models directly from the fragmentfile.
Any suggestions are welcome.
I think maybe a simple perl script can convert this into a pdb files like the NMR structure format.
The fragment file does not contain enough information to make the fragments you want to see. I don't remember what all the columns are, but there's not enough of them to generate the PDB fragment. Here's what's present:
"1fp5 A 146 P L -56.857 -172.981 -176.608 2.834 12.355 25.732 3 0.000 P 1 F 1"
PDB chain resid sequence SS phi psi omega ? ? ? ? ? ? Frame Frame#
Some of the ? are related to quality-of-fragment versus the SS info provided.
Anyway, there's not coordinates there for the sidechain, and it assumes ideal bond lengths and angles (so only backbone dihedrals are provided). If you want to look at the PDB, then you'll need to extract as you're doing. Otherwise I guess you could generate idealized fragments, but it would be fairly complicated, as you'd need to convert internal coordinates to XYZ. I don't think Rosetta has a method for dumping a fragment set as a PDB or series of PDBS; I'll ask.
Finally: fragments are not supposed to bridge unknown regions of partially built structures - if you have a mostly solved core and unknown loops, the fragment picker doesn't pick loop fragments with endpoints to close that gap (it doesn't know what the gap is). It is possible to choose fragments for this purpose, but code to do that is unreleased.