INTRODUCTION
PREDICTION OF HIV-1 PROTEASE/INHIBITOR AFFINITY USING
ROSETTA LIGAND DOCKING WITH FLEXIBLE XML PROTOCOLS
TOWARDS LIGAND DOCKING INCLUDING EXPLICIT INTERFACE WATER
CONCLUSION & FUTURE DIRECTIONS
Experimental ΔΔG & Ki values for HIV-1 PR/PI binding
Template structures used for comparative modeling
Sequence Alignment of 171 HIV-1 PR backbone templates
HIV-1 PR/PI data partitioned by location of mutations
Distribution of Rosetta energy scores for a set of HIV-1 PR/PI models
Comparison of top Rosetta models and experimental structures
HIV-1 PR/PI docking protocol
Until the final minimization, the repulsive score term is downweighted so that small collisions are allowed (soft_rep). The final minimization includes minimization of backbone φ/ψ angles (minimize_backbone) with harmonic constraints on the Cα atoms, where 0.2 Å is a standard deviation (harmonic_Calphas 0.2).
Prediction of clinical outcomes
Ligand docking through incremental construction
The figure below (next page) represents our vision for improved ligand flexibility modeling within RosettaLigand. The problem is reduced by using pre-generated conformations based on rotation profiles for each atom-type pairing found in the Cambridge Structural Database (Kaufmann, Glab et al. 2008). By splitting a ligand into multiple fragments, creating conformers for each fragment, and fitting fragments one at a time, the competing goals of flexibility and efficiency can be balanced.
Now residues can connect to any number of other residues, they can connect to the same residue through multiple bonds, and they can connect to different residues through the same atom. As a tool originally constructed for protein folding, the Rosetta code included the assumption that most residues, as part of a peptide chain, would connect with two other residues, upstream and downstream, and a few would connect with 3 residues (including a disulfide bond). . To perform fast and efficient rotamer sampling, rotamer energies are stored in a “trie” data structure for rapid retrieval (Leaver-Fay, Kuhlman et al. 2005).
The maximum common subgraph algorithm (Shen, Lange et al. 2008) was used to search for fragments within larger molecules. This work was performed using BCL, a chemical informatics software package developed in the Meiler laboratory. These entries can simply be PDB files containing every fragment instance found in the CSD.
Alternatively, rotamer libraries can be represented by files listing the combinations of torsion angles for each rotatable bond, along with the propensity of those sets to torsion within the CSD. Sets of torsions that occur more frequently in the PDB may acquire more favorable Rosetta energies. Future work includes writing an algorithm that links an initial fragment to its conformers, and then links additional fragments via multiple rounds of ligand conformer docking (see pseudocode below).
34;Lessons in molecular recognition: The effects of ligand and protein flexibility on molecular docking accuracy." Journal of Medicinal Chemistry. 34;Automated flexible ligand docking method and its application for database searching." Journal of Computational Chemistry. 34;Consistent blind protein structure generation from NMR chemical shift data." Proceedings of the National Academy of Sciences of the United States of America.
Ligand design through incremental construction
(3) specifies growth termination criteria, including molecular weight, number of hydrogen bond acceptors, number of hydrogen bond donors, number of heavy atoms, or total number of atoms; With these elements, we tried to implement the following algorithm, presented as pseudo code… .. create a fragment library with conformers for each fragment by fragment in the fragment library .. dock fragments sample fragment conformers and side chain rotamers predict binding affinity . keep the strongest binding affinity fragment as the starting fragment while user-defined growth limits are not reached .. connect a random fragment at a random link point sample fragment conformers and side chain rotamers sample the rigid body position for the extended molecule .. accept or reject new growth using The Monteja Carlo approach minimizes backbone, side chain, and ligand torsion angles. The mol_to_params.py script has been modified to allow the creation of these small molecule fragments.
The creation of small molecule fragments begins with the identification of MOL or MOL2/MDL files of molecules that contain the fragments that the user wishes to add to his/her fragment library. Ligand design begins with the docking of an initial molecule fragment from a library of fragments. The HeavyAtom filter terminates growth when the number of non-hydrogen atoms has reached a cutoff value, while the AtomCount filter relies on the total number of atoms, incl.
Similarly, the MolecularMass filter and MolarMass filters stop the expansion of small molecules when their limits are exceeded. The ChainExistsFilter is useful to ensure that the ligand design occurs only for the starting fragments positioned in the binding pocket. After termination of growth of a small molecule, the molecule is likely to contain junctions that are not connected to other small molecule fragments.
Described below is a complete XML algorithm for designing a small molecule fragment-by-fragment using the new Rosetta code. Grow and dock in a loop until clipping filters are hit
XML used with standard docking (chapter 4)
XML used with protein-centric docking
XML used with ligand-centric docking
Summary of my commits to the Rosetta SVN server
'n Mini/test/scientific/cluster/ligand_dock_scripts/translate_rotate.xml M mini/src/protocols/ligand_docking/MinimizeBackbone.cc. 'n Mini/test/scientific/cluster/ligand_dock_scripts/native/1ppc_1pph.pdb.gz D mini/src/protocols/ligand_docking/LigandArea.hh. 'n Mini/test/scientific/cluster/ligand_dock_scripts/input/1dm2.params M mini/src/protocols/ligand_docking/MinimizeBackbone.hh.
D mini/src/protocols/ligand_docking/LigandArea.cc Një mini/test/integration/tests/ligand_dock_script/flags. Një mini/test/scientific/cluster/ligand_dock_scripts/input/1p8d_1pq6.pdb.gz Një mini/src/protocols/ligand_docking/InterfaceScoreCalculatorCreator.hh. M mini/test/integration/tests/multi_residue_ligand_dock/flags M mini/src/protocols/ligand_docking/Rotate.cc.
M mini/src/protocols/ligand_docking/ligand_options/chain_functions.hh M mini/test/integration/tests/multi_residue_ligand_dock/flags. M mini/test/profile/tests/multi_residue_ligand_dock/ligand_options.txt A mini/src/protocols/ligand_docking/RandomConformersCreator.hh. M mini/test/scientific/cluster/multi_residue_ligand_docking/submit.py M mini/src/protocols/ligand_docking/GrowLigand.hh.
