Gilles Truan wrote:
I created 5 models with a modified automodel script:
...
These models have their OBJECTIVE FUNCTION around 11,000 . I made some
DOPE plots to check for poorly defined regions. I detected some loops
where the DOPE score was higher. I decided to further modelize these
loops with loop_model as shown in the tutorial selecting residues having
DOPE score significatively higher than the average DOPE for the entire
structure. Here is the script:
class MyLoop(loopmodel):
def select_loop_atoms(self):
return selection( self.residue_range(1, 26),
self.residue_range(75,84),
self.residue_range(89,97),
self.residue_range(102,122),
self.residue_range(183,204),
self.residue_range(230,238))
...
Quite surprisingly, the OBJECTIVE function of the 25 produced loop models varies from 2000 to 7000 and the DOPE plots of the different loop models is not as good as the first models!!
You can't compare the objective function for model building with that for loop modeling, since the form of the function is different (loop modeling only applies the potential to part of the protein - the loops - and it uses a statistical potential which is not used in model building).
Is there something wrong in the script or do I need to change and optimize some other parameters?
It is extremely unlikely that you will be able to effectively optimize your loops, since you are trying to optimize several extremely long loops simultaneously. Each loop should be no more than 12 residues long.
Ben Webb, Modeller Caretaker
--
http://www.salilab.org/modeller/
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage