Hi Tom,
I wonder why it happens that an experimentally established disulfide bond is not confirmed in several crystal structures. Maybe the crystallization condition were not oxidative to support disulfide bridge formation. In this modeling case the template selection step must be crucial: a structure with a disulfide bridge, even if the structure itself is not the highest resolution one available, should enjoy priority.
If you do not have such template at all and if you believe that for your question the physiologically meaningful situation is an oxidized Cys connection, I would suggest to patch it from the beginning. You can ask for a simultaneous, exhaustive MD/SA refinement. I believe it will be more accurate when all the restraints used by Modeller will be taken into account all the way during optimization (packing). To set the highest level of MD/SA optimization include in your TOP file
SET MD_LEVEL='refine1'
best wishes,
Andras
Tom Duncan wrote: > > There are several good template structures for our target protein but, > although it is known that a disulfide can form between a specific pair > of cysteines in theses proteins (cysteines conserved in templates & > target), the known structures show these cyteines as 15-20 angstroms > apart. Nearby flexible loops or "hinge" regions can be identified. Would > it be reasonable to model the target directly with a disulfide patch, or > better to obtain an initial model based on the templates and then do > more thorough MD/SA cycles with a restricted region of the model and S-S > distance restraints before final refinement with the disulfide patch? > > Thanks, > > Tom Duncan > -- > Thomas M Duncan > Research Associate Professor > Dept Biochemistry & Molecular Biology > Institute for Human Performance, Rm 4311 > SUNY Upstate Medical University > 750 E Adams St > Syracuse, NY 13210