Hi Andrew, Yes, translating 3J couplings into rough torsion angle restraints is standard practice. Anything to remove false minima is a good idea. Note that Gabriel Cornilescu's TALOS can often do this automatically for you for the backbone torsions, since TALOS's predictions are right 98% of the time.
The advantage of including the 3J couplings directly as restraints is in cleaning up the final torsion angles as much as possible. It's a refinement tool, not a structure determination tool. And since very small systems often have very few NOEs, pushing your 3J information as far as it can go is obviously valuable. --JK On Aug 15, 2007, at 9:56 PM, Andrew Borgert wrote: > Thanks for the input, it was very helpful. > > I'd like to extend this discussion a bit if that's alright. I've > noticed in several papers that people translate j-coupling values into > dihedral restraints in an effort (presumably) to avoid the problem of > multiple karplus solutions. For example, a coupling greater than 8 Hz > is translated to a phi dihedral of -140 +/- 40, even though +140 is > also > a valid solution to the karplus equation (granted that +140 is out in > the no-mans land of the Ramachandran plot). Is this sort of thing > considered standard practice? Is it even worth doing on a small > system > (15 residues)? > > Thanks > Andrew Borgert
