On 07/06/2012 10:36 AM, Adhikari, Badri (MU-Student) wrote: > I am using edited version of > http://www.salilab.org/~drussel/unstructured/imp_doc/atom_examples.html#stru... > to generate structures from amino-acid sequences. This example uses > IMP.atom.CHARMMTopology. > > I would like to feed contact information as a restrain to guide the > prediction process.

The script you are running is not really "predicting" 3D structures - they are generated purely from internal coordinates - so you'll always get extended chain structures (no secondary structure) with no consideration of excluded volume (clashes) or folding.

> 1.What should be the approach? Which classes/functions should be > used?

You can use the extended chain structure as an input to an actual prediction, assuming you can express your contact information as one or more spatial restraints. Add stereochemistry, excluded volume, and your contact information restraints to the scoring function, and sample with the method of your choice (e.g. CG, MD, MC). There is a charmm_forcefield.py example which demonstrates the setup of the restraints, and there are several examples of optimizers.

Ben