Ivan Anishchenko wrote: > I deal with a peptide (35 residues long). I know its amino acid sequence > and a set of mean values for phi- and psi- dihedrals and the respective > confidence interval for each dihedral angle. The data is obtained from > the NMR experiments. The aim is to obtain a structure with the > main-chain dihedrals lying within the corresponding confidence interval > for every residue. Side chains can take any conformation. How can such > restraints be described in MODELLER?
1. Read in the initial structure (see the manual for complete_pdb) or build an extended chain from the AA sequence (model.build_sequence).
2. Add restraints on each dihedral. See Restraints.add. To get the list of atoms for each dihedral type in each residue, see Residue.phi and Dihedral.atoms. For example, to constrain every phi angle in the model 'm' to 0 +/- 0.1 radians, using a Gaussian restraint:
for res in m.residues: if res.phi is not None: atoms = res.phi.atoms f = forms.gaussian(group=physical.dihedral, feature=features.dihedral(*atoms), mean=0, stdev=0.1) m.restraints.add(f)
3. You may also want to build other sets of restraints, for example stereochemical restraints. See Restraints.make.
4. Optimize the structure. See conjugate_gradients().
Ben Webb, Modeller Caretaker