Domain orientation in two-domain modelling
Hello,
I have been running some experiments attempting to combine two separate domains into a single chain model. To do this, I have the templates of the two domains I am interested in, and align the the sequence to the respective domains. I am then using restraints to bring the two domains together, attempting to replicate the native interface.
However, after initial runs of Modeller, it was clear that the starting domain orientations drastically impacted the resultant models, which appears to be in line with the starting orientation used: http://salilab.org/archives/modeller_usage/2007/msg00039.html , point 1.
After randomising the starting orientations of the two separate domains, in some instances the models can be minimised well, satisfying the added restraints, whereas for other models in unfavourable orientations, these models cannot be minimised and only a few restraints are being satisfied.
Do you have any suggestions for how to get Modeller to re-orientate the domains in these cases where they have unfavourable starting orientations?
Thank you.
On 7/6/15 8:34 AM, Stuart TETCHNER wrote: > After randomising the starting orientations of the two separate domains, > in some instances the models can be minimised well, satisfying the added > restraints, whereas for other models in unfavourable orientations, these > models cannot be minimised and only a few restraints are being satisfied. > > Do you have any suggestions for how to get Modeller to re-orientate the > domains in these cases where they have unfavourable starting orientations?
Presumably in the unfavorable case the domains overlap or are entangled in some way so that Modeller can't get them unstuck. I would expect the energy to be pretty high in this case. So you should be able to just throw away high-scoring models and do reasonably well.
Another option would be to start with the two domains widely separated, so there is no chance of any steric clashes occurring. During optimization Modeller should bring them together in order to satisfy your interdomain restraints but (hopefully) should not introduce any new clashes.
Ben Webb, Modeller Caretaker
participants (2)
-
Modeller Caretaker -
Stuart TETCHNER