I am trying to perform an abinitio modelling of the full-length structure of a two domains protein, having the crystal structure of the Nterminal one. If I try to use a similar protocol as the demo of comparative modelling with end extension the program stops with the following error:
ERROR: moving_connection.icoor().is_internal() && fixed_connection.icoor().is_internal()
ERROR:: Exit from: src/core/conformation/util.cc line: 100
sometimes the first pdb that comes out is ok, but when it crashes, the position of both domains is strangely far away.
Thanks in advance!
This isn't really enough evidence to help us debug the problem. I can tell you that in general, this is an error that occurs when adding residues to a pose; probably it is occurring when Rosetta is tacking the unknown-portion residues onto the known-portion residues.
My best guess is a misalignment or misnumbering in the input files; for example your alignment inputs tell Rosetta that there are 100 residues in the N-terminal domain, but it only contains 96 readable residues (perhaps because of nonzero occupancies, or undetected funny numbering). This guess is a little inconsistent with the "sometimes it crashes, sometimes it works but wrong" aspect, but very consistent with what historically causes problems for this sort of modeling.
thanks for the response
I checked the pdb and I didn't find anything unusual
the broker.arg is identical as the demo and the files I am using are almost the same:
RIGID 1 126 0 0 0
the only different thing is that I am trying to model the C-terminal domain. I also saw the following in the output:
protocols.general_abinitio: (0) AbrelaxMover: S_0001
loops: (0) Extended: (-3,+0) LOOP 124 326 0 0 0
protocols.topo_broker: (0) Enlarged loops:
protocols.topo_broker: (0) LOOP begin end cut skip_rate extended
protocols.topo_broker: (0) LOOP 124 326 0 0 0
this is not right because the modelling part is between the residue 127 and the end (position 199). What I found completely weird is that the first cycle worked fine, with a pdb at the end with 199 residues all connected. But in the second cycle the error appears and both domains are too far away from each other.
sorry...I solved the problem. I didn't realize that I have to start with the full-length protein as a model
Glad to hear it!