On 04/14/2010 08:08 PM, Song, Hyundeok (songhk) wrote: > After alignment, I used loop patch to generate the model. I thought > using loop modelling gives more accurate structure for homology > modelling.
No, loop modeling will probably give better structures for the loop regions than regular homology modeling, since homology modeling doesn't do any refinement of the loops. But it won't affect the aligned regions at all.
> When I checked model pdb structure, it gave the warning > message "unusual bond length, bond angle, Chirality deviations". I > think this error comes from 70 fasta sequence length difference > between model and template as suggested from Modeller Caretaker.
Your original question suggested you had a single 70-residue long loop. Looking at your alignment, I now see that that's not the case - you have several shorter loops. Thus there aren't really any parts of the template you can remove.
> My question: In order to generate best model from Modeller, which > residues of template should I delete?
It depends on what you want to use the model for. It may be that the warning you're getting can be ignored. Or you have several loops in your structure, so you should concentrate your loop refinement on just the loops that you are most interested in (for example, loops that are close to a ligand binding site or an interface region). Just doing default loop modeling will try to simultaneously refine all of the loops.
Ben Webb, Modeller Caretaker