On 09/07/2015 11:27 PM, thereal sisterdot wrote: > I decided to try usingupper_bound restraints as above to feed the > crosslink information to modeller. > > But actually the restraints did not improve the conflicting distances. > my assumption is that the few restrains from the crosslinks can not > compete with the many restrains coming from the alignment. > I guess one would have to modify the model later somehow.
Modeller will always build a model that resembles the template (in aligned regions) by construction - that's what it's designed to do. So in order to satisfy your crosslinks you would have to remove the corresponding distances from the template. One way to do this would be to break the template into multiple pieces (e.g. one per chain or domain), i.e. changing your alignment from
template AAAAA/BBBBB model mmmmm/mmmmm
to
template1 AAAAA/----- template2 -----/BBBBB model mmmmm/mmmmm
Then you'll get no A-B interactions enforced in your model because those chains are in different templates.
> maybe using imp?
Sure, there are many successful applications of crosslinking using IMP. See http://integrativemodeling.org/systems/?tag=chemical%20crosslinks for recent published examples.
Ben Webb, Modeller Caretaker