On Wed, 24 Mar 2004 11:56:16 +0000
Binbin Liu <> wrote:
> Does anybody know why. Does MODELLER calculate the torsion
> based on some rules or it is by chance.
Modeller calculates each structure by an algorithm that uses
random numbers to search for the best positions for each
part of the molecule. This means that each structure will differ
slightly, even though the same data (the templates) were used as input
to the algorithm. This is why you see the variation - its not really by
chance though. There are energy functions involved in side-chain
placement, and a final optimisation stage that moves them around so
they don't touch too much.
Side chain torsion angles are actually quite difficult to get right -
they are the 'fine detail' of structure, and making them perfect means
having a highly accurate model of the system. There are tools
dedicated to modelling side-chains efficiently - search for SCWRL by
Roland Dunbrack at the fox chase CRC, for one example. Even then, if the
model of the backbone is not perfect, then the side-chains are not
likely to be 100% correct.
j.
_______________________________________________________________________
Dr JB Procter:Biomolecular Modelling at ZBH - Center for Bioinformatics
Hamburg http://www.zbh.uni-hamburg.de/staff.php