Michel Seigneuret wrote: > Here is my problem. I would like to build a model of a 2 domain > protein. I have templates for each domain as well as additional > interdomain distance restraints from crosslinking data. > As I understand it is possible to introduce these additionnal distance > restraints in the input file.

Certainly - just define them in the special_restraints routine. See http://salilab.org/modeller/manual/node26.html

> However I also notice from FAQ n°1 that in general :"...the relative > orientation of the two domains the resulting model will probably have > an incorrect relative orientation of the two domains ...". > So I suspect there will be a conflict between my (correct) additionnal > interdomain distance restraints and the (incorrect) interdomain distance > restraints resulting from the initial arbitrary relative orientation of the > 2 templates. > How can I solve this. Is there a way to tell Modeller not to use these latter > restraints or a way to remove these at some point ?

It's not a problem. Modeller does not add inter-template restraints, since there is no data there. If as you say each template covers a single domain, and there is no overlap between them, then Modeller cannot know anything about the interdomain interaction, and thus any model you build is likely to be incorrect, as is stated in the FAQ. But this is due to the absence of any interdomain restraints, not the imposition of incorrect ones. So you don't need to worry about it in your case - simply add your correct restraints.

(BTW, if you did want to remove existing restraints, you can either manually edit the .rsr file and then use that as your custom restraints file - see http://salilab.org/modeller/manual/node24.html - or you can use the restraints.unpick() command followed by restraints.condense().)

Ben Webb, Modeller Caretaker