You are here

Ligand docking and protein design of an internal pocket in a protein with a cofactor

3 posts / 0 new
Last post
Ligand docking and protein design of an internal pocket in a protein with a cofactor

Hi, I am a brand new user of Rosetta... I want to model in different substrates in a transporter, and see how the substrates affect the overall protein structure. The problem is, the binding pocket I am trying to look at is pretty much in the center of the protein, so one question is: how can I define a region of the protein where I want my ligand to dock? A confounding variable is that my protein also has a cofactor covalently bound to a residue side chain. A colleague in my lab tried to help me, but we got stuck with this covalently bound cofactor issue. Could someone please give us a strategy of how to deal with the cofactor in the PDB file, as well as how to deal with the docking area being a pocket inside the protein?

Thanks in advance for any help!

Post Situation: 
Mon, 2011-06-20 13:55

The standard ligand_dock application ( allows the definition of various pockets through the -start_from and -uniform_trans flags. If you just have one pocket, I believe you can simply start with a structure with the ligand in the center of the pocket, omit the -start_from flag, and then specify the size of the pocket with the -uniform_trans flag. (I tend to use the to set things up, and it takes care of most things automatically.)

The covalently bound cofactor is a little more of an issue. Although I've never used ligand docking with multiple non-protein ligands, I believe that the non-docked ligand isn't allowed rigid body movement. I would suggest first trying to model the covalent cofactor as a non-covalent additional ligand. You may have a bit of an issue with the residue wanting to repack to relieve the steric clash (ligand docking always keeps the backbone fixed), but if the attachment point is far enough from the ligand docking site, it might not be an issue. Alternatively, you just remove some of the ligand atoms. If those methods won't work for you (e.g. if the attachment point is involved in ligand binding), you could try modeling the covalent cofactor as a non-canonical amino acid, but that can be a bit tricky, and it's easiest to just avoid it.

Although not mentioned in the body, you also mentioned protein design in your subject line. My recommendation is that if you're looking for mutations, you should use the enzyme_design application ( Despite the name, it should work for ligand binder design in addition to enzyme design.

Mon, 2011-06-20 20:06

in case you haven't found a satisfying solution, i'd suggest using an enzdes style constraint file to specify the cofactor as covalently bound. check out the enzdes cstfile documentation at
i'd make a cstfile containing one block, which specifies a constraint between the two atoms that are covalently bound. the force constants can be set to 0, but the constraint block needs to be declared as covalent.
once you have the constraint file, simply reference it one the command line with -enzdes:cstfile

Tue, 2011-07-05 18:46