Protein docking is an important computational technique for understanding protein-protein interactions. It seeks to predict the binding orientation of two protein structures. Docking is a difficult problem due to the large search space of possible orientations and conformations. Successful docking requires scoring potential poses based on factors like free energy, buried surface area, hydrogen bonding, and complementarity of charges. Several search strategies are used like simulated annealing and genetic algorithms. International blind tests like CAPRI evaluate docking methods and provide lessons on better accounting for flexibility and active sites. Future challenges include improving scoring functions and models of flexibility.

1. Protein docking
Saramita De Chakravarti
12/14/2012

2. Outline
 Physical background of protein
 Protein docking

3. Quantum Mechanism (QM)
 Electrons
 Nuclei
 Wave functions instead of Newton Eq.
 High computation cost
 Ab initio N4
N is # of base functions
 Semi-empirical N3
 Molecule of a couple of hundreds atoms
 Pico-second(10-12
s) behavior

4. Molecular Mechanism (MM)
 Atoms
 Newton equations
 Parameters from experiment and QM
 force field
 System with 100,000 atoms
 µs (10-6
s) behavior

5. Force field
+−+−= ∑∑ 22
)()( eq
angle
eq
bonds
b krrkU θθθ
])[()]cos(1[
2 612
ij
ji
ij
ij
i ij ij
ij
dihedral n
n
r
qq
r
B
r
A
n
V
⋅
+−+−+ ∑∑∑ ∑ > ε
γφ
kb – bond parameter
kө --- angle parameter
Vn --- dihedral energy barrier
Van Der Waal radii
Partial charge set

6. Solvent effect
 Oil in water
 Hydrophobic interactions
 Alcohol in water
 Hydrophilic interactions
 Explicit water
 Solvent Accessible Surface
 Surface Complementarity

7. Twenty Amino Acids

8. Sidechains
The amino acids vary in their side chains (indicated in blue in the diagram).
The eight amino acids in the orange area are nonpolar and hydrophobic.
The other amino acids are polar and hydrophilic ("water loving").
The two amino acids in the magenta box are acidic ("carboxy" group in the side chain).
The three amino acids in the light blue box are basic ("amine" group in the side chain).

9. Peptide bonds

10. Proteins
 Primary structure
 Secondary structure
 Tertiary structure
 Quaternary structure

11. Important forces
 Weak forces
 Delicate balance
 Electrostatic (hydrogen bonds, salt
bridges)
 Hydrophobic (sidechain packing)

12. Protein docking

13. Contents
 Why Is Docking Important?
 Why Is Docking Hard?
 Docking Scoring Criteria
 Docking Search Strategies
 CAPRI Participants & Algorithms
 Lessons Learnt From CAPRI
 Selected Docking References
http://www.csd.abdn.ac.uk/~dritchie/

14. Why Is Docking Important?
 Better understand the Machinery of Life
 Enzyme-inhibitor class
 Antibody-antigen class
 Others
 Engineered Protein Enzymes
 Protein Therapies
 Drug targets

15. Why Is Docking A Hard Problem?
 Large Search Space !
 Rigid Body Docking: 6D
 Flexible Docking: 3N (N normal modes)
 Structure Space: Continuous

16. Criteria for Good Docking
Orientations
 Low Free Energy
 Low Pseudo-Energy Based On force field
 Large Surface Burial
 Small van der Waals Overlaps
 Good H-Bonding
 Good Charge Complementarity
 Polar/Polar Contacts Favoured
 Polar/Non-Polar Contacts Disfavoured

17. Docking Search Strategies
 Pseudo Random
 Simulated Annealing / Monte Carlo
 Genetic Algorithms
 Directed Search
 Geometric Hashing
 Spherical Harmonic Surface Triangles

18. Docking Search Strategies
 Brute-Force Search
 Explicit Grid Correlations
 Fast Fourier Transform (FFT) Correlations
 Refinement Phase
 Classical or Soft Potentials (+/- Electrostatics)
 Desolvation, Solvent Dipoles...
 Visual Inspection!!

19. CAPRI
 Critical Assessment of PRedicted
Interactions
 http://capri.ebi.ac.uk/
 targets available (unbounded/bounded)
 Bounded (rigid)
 Unbounded (non-rigid)
 International groups participated

21. Cartesian Grid Correlations
Basic Principles

24. d
PNAS 1992 89 pp. 2195-2199

25. Electrostatic complementarity
JMB 1997 272, pp 106-120

26. The electric field by protein A

27. Correlation function
Binary filter to remove false positive geometries

28. Lessons Learnt From CAPRI
 Antibodies Often Bind Near Antigen's
Active Site
 Expect Large Conformational Change
in Enzyme/Inhibitor Docking
 Develop Better Models of Flexibility

29. Future Challenges For
Docking
 Better Scoring Functions
 High-Throughput Screening
 Tractable Models of Flexibility

30. Reference
 Reviews
Halperin et al.; Proteins, 47 409-443 (2002)
Smith & Sternberg; COSB, 12 28-35 (2002)
 CAPRI
http://capri.ebi.ac.uk/
 Algorithms
Chen & Weng; Proteins, 47 281-294 (2002)
Fernandez-Recio et al.; Prot Sci, 11 280-291 (2002)
Gardiner et al.; Proteins, 44 44-56 (2001)
Camacho et al.; Proteins, 40 525-537 (2000)
Palma et al.; Proteins, 39 372-384 (2000)
Ritchie & Kemp; Proteins, 39 178-194 (2000)
Gabb et al.; J Mol Biol 272 106-120 (1997)
Vakser; Proteins, S1 226-230 (1997)
Abagyan et al.; J Comp Chem, 15 488-506 (1994)
Norel et al.; Prot Eng 7 39-46 (1994)
Katchalski-Katzir et al.; PNAS, 89 2195-2199 (1992)