Library For Generating Protein Conformations

Question

[Flow]

I have some protein PDB files, and I would like to generate new conformations, just changing the values of some dihedral angles. There is any library for this?

PS: I would also be interested in a CPU or GPU computationally efficient library. I mean, pymol can do this, but think about millions of conformations.

[Flow]

ok, self answer: http://www.pymolwiki.org/index.php/Set_Dihedral

[Hans Laxness]

you could also generate them setting up a simulation in gromacs

Answer 1

[dimkal]

you should look at Montecarlo simulation. One of my favorite MC method is MMC (by M. Mezei) which is very efficient way of sampling different conformations.

http://inka.mssm.edu/~mezei/mmc/

[Flow]

I want to do my own sampling stuff, but do not want to code basic things like dihedrals, etc, that is why I am looking only for conformation generations library

[dimkal]

Then I would suggest for you to look into a software package called Tinger. It also has MC-based sampling but it's written in such a way that it's easily (relatively) hackable. There you can probably define your own sampling schemes. (http://dasher.wustl.edu/ffe)

Answer 2

[Keith Callenberg]

You might take a look at CRANKITE. There is a paper describing how to use it here.

It has a sub-program called peptide:

"The Metropolis polypeptide backbone conformation sampler. A very short simulation can be run by using a command like:

./peptide -r 1000x25 -t 201 d1ctf__.pdb

This will read the initial conformation from file d1ctf__.pdb, perform 25,000 Metropolis moves, while recording a contact map and a snapshot conformation every 1000 steps. The output will have 25 contact maps and 25 snapshots."

Another user recommended crankite to me when I was looking for a tool to toggle single rotamers (not randomly changing the whole structure), but it may be closer to what you are looking for.