RosettaDock-5.0 server documentation

Summary: The RosettaDock 4.0 Server performs a local protein-protein docking search with backbone ensembles. The ensemble docking protocol is based on the conformational-selection mechanism of protein docking. It utilizes an ensemble of the protein partners pre-generated with Rosetta Relax, Backrub and NMA to sample diverse backbone conformations while docking. In the low resolution stage, each docking run performs rigid-body translation and rotation around the protein partner while swapping backbones from the pre-generated ensemble. This allows the sampling of diverse backbone conformations while docking. In the high-resolution stage, an all-atomistic refinement is performed over the generated encounter complex and side-chains at the interface are packed for optimal binding. The output decoy is then evaluated with Rosetta score function and the best binding complex structure is determined on the basis of Interface Score. Note, the server will only work for protein docking cases. For ligand-protein, membrane-protein or glycan-protein docking, please follow respective scripts in literature.

1. You must upload a reasonable guess for the starting position. Place the protein partners near contact (but not overlapping) with the relevant patches of the proteins facing each other.

2. PyMol can be a useful tool to position proteins relative to each other. Use "editing" mode from the right panel, and try right-clicking to select a chain and "drag" to enable translation and rotation of the molecule (typically requiring left-shift + middle and left buttons). Finally from the main menu, File→Export Molecule can be used to write a PDB file containing the starting structure with both docking partners. The ‘align’ command may also be helpful if you are using a homologous complex as a guide.

   Alternately, starting positions can be creating using one of several docking servers which perform global searches. Some leading servers include ClusPro, GRAMM-X, HEX, PatchDock, SymmDock, ZDOCK. Note that coordinate file formats from these servers might need to be brought into compliance for use by our server (e.g., putting a TER record between docking partners, assuring that the occupancy field is present and standard atom and residue names are used, etc.).

3. Docking partners can be uploaded as two separate PDB files. If there are multiple chains which serve as a single docking partner (e.g., the heavy and light chains of an antibody), create the protein partners such that the each partner represents a unit that needs to be docked. For e.g., we dock antibody (chains H and L) with an antigen (chains A and B), then the docking partners would be partner1 (chain H,L) and partner2 (chain A,B). Further, make changes in the partners file to denote the docking partners so that fold tree could be re-assigned. In this case, the partners file will have the partners as AB_HL (where the underscore indicates chian break). Further description on the PDB file format description can be found here.


4. Given the local nature of the search, there is no need to include extra domains of the proteins beyond the two interacting domains. Trim unneeded residues out of your PDB file before uploading. The server will not accept PDBs larger than 600 residues total.
   
   
5. RosettaDock 4.0 follows an ensemble generation step followed by ensemble docking. First, the input protein partners are used to generate an ensemble of 100 structures each, with diverse backbone conformations using Rosetta Relax (30 structures), Backrub (30 structures) and NMA (40 structures) respectively. Note that the ensembles are generated independent of the protein partner. After the ensembles are generated, we prepack the structures and initiate local docking.
   
   
6. Local docking performs perturbation of ~ ±3 Å in the direction between the two proteins, ~8 Å in the directions sliding the proteins relative to each other along their surfaces, ~8° of tilt of the proteins, and a complete 360° spin around the axis between the centers of the two proteins. The server will start 1000 independent simulations from this range of random positions. Swaps are also performed with conformations from the pre-generated ensemble to accommodate for backbone flexibility in docking.