Hallo Modellers!
I try to model protein binding sites including ligand information in terms
of distance dependent atom pair potentials.
For testing this approach I tried to predict the orientations of binding
site residues. After generating 100 models of a protein I compare them
with the actual crystal structures. The predictions are quite reasonable.
However, there seem to be some problems in the side chain conformations of
the aromatic amino acids (maybe because of steric hinderance). In some
cases there is only one preffered conformation of, e.g. a tyrosine, i.e.
only one preffered value of the CHI1 dihedral angle.
I have read in the manual that there is the possibility to
ROTATE_DIHEDRALS by randomizing. How do I have to set up my top file to
tell MODELLER to select certain amino acids (e.g. by number) and to
randomize the CHI1 and/or CHI2 angles and - if possible - to generate
models with diverse CHI1 and/or CHI2 angles, even if some are not
energetically favourable? The rest of the protein should be modelled as
usual.
Do you think this procedure an advisable way to do or do you have any
better idea how I can constrain the modelling process to generate more
diverse dihedral angles for selected amino acids?
Thank you for your answer,
nice regards,
Andreas
--
######################################################################
# Andreas Evers #
# #
# Research group for Drug Design & X-ray Crystallography #
# Institute of Pharmaceutical Chemistry #
# Philipps-University of Marburg #
# Marbacher Weg 6 phone2 +49-6421-28-25072 #
# D-35032 Marburg FAX: +49-6421-28-28994 #
# email: #
# AG-Klebe-Home: http://www.agklebe.de #
######################################################################