Docking is a an approach to predicting how molecules bind together. Given the fixed chemical sequences and initial conformations of the individual partners as input, rigid body alignment, sidechain and backbone degrees of freedom are sampled to search for bound conformations that have favorable over all energy.
As a scientific benchmark, accurately predicting experimentally validated binding interactions is a stringent test of the energy function and conformational sampling. Depending upon the level of detail of the input data and how successful predictions are evaluated, the docking scientific benchmark can emphasize the energy function or the conformational sampling.
The docking local refine scientific benchmark emphasizes evaluation of the energy function by providing the experimentally validated bound conformation of both interaction partners as input and limiting the allowed confomational sampling. A physically realisitic score function should recovery the native conformation a substantial fraction of the time. Since sampling is not required to recover the native conformation, any significant deviation from the native conformation indicates that configuration molecular interactions in the native conformation is inappropriately disfavored relative to alternative binding conformations.
For documentation about how to run the docking_protocol application see docking protocol
The docking scientific benchmark uses the ZDOCK benchmark v4 set.
The docking local refine benchmark starts with the native bound receptor and ligand conformations.
Compile Rosetta
cd mini_base_dir
./scons.py bin mode=release -j<num_processors>
Run test
cd mini_base_dir/test/scientific
./scientific.py docking_local_refine -d mini_database_dir
Investigate results
cd main/tests/scientific/cluster/docking_local_refine/outputs