Subject: Re: [modeller_usage] Questions on loop modelling
From: Modeller Caretaker <>
Date: Fri, 16 Jun 2006 14:58:40 -0700
Cc:
hori koshii wrote:
1. Since 12aa is the limit of loop modelling, is that viable to build a
"model" with the "template sequence plus 2 aa of the "target"
insertion, then use this "template +2aa" model as template for building
the model of the target????
There is no 'limit' to loop modeling, but the longer the loop, the
larger the space that must be sampled. Your approach doesn't solve the
problem, because you still have to sample a 14aa loop, whether you do it
all at once or in chunks. You can certainly use loop modeling for 14aa
insertions, but you should be aware that the predictions aren't as
reliable as those for a shorter insertion.
2. The residues flanking the insertions also belong to the loop regions
(of the template), so, my question is, when we do loop modelling, do we
need to model the whole piece of loop for accuracy??
If you think they need to move, you need to include them. Modeller uses
the residues immediately adjacent to the loop region to anchor the loop
(this reduces the space that must be searched) and so does not move them.
3. When we generate loop models, do you model one loop and then the
other? Since I have two loop regions, I wonder if I should build (~500)
models for one regions and then another set for the other region? Does
one loop region actually affect the conformation of the other loop? Do
you think it's necessary to build them both at the same time?
That depends on how close the loops are in space. If they are far apart,
you can certainly model them both at once.