2. Clustering pose

command: cluster.linuxgccrelease @flags-cluster > cluster.log

Clustering was used for largest population pose recognition which usually represent the most likely protein binding mode. The best scored pose in the largest clustered group would be used as the input of next step.

3. Initial pose refinement

command: relax.linuxgccrelease @flags >relax.log

Initial docking in 1st step is a rigid body docking without sidechain and the 2nd would build full atom poses. There are many unfavorable sidechain conformation or atom clashes in the poses comes from step 2. So this step would make the pose much more favorable in energy level.

4. High resolution docking

command: docking_protocol.linuxgccrelease @flags2 > docking2.log

To confirm the docking results, small perturbations can be introduced for such purpose. In this step, thousands of perturbed pose can be generated and the best one can be identified as final docking results according to the scoring function in Rosetta.

Tips:
Some protocols in Rosetta could support parallel job running, users could use MPI for such kind tasks so that the whole work could be finished much faster.