Notes

Slide Show

Outline

1	Algorithms for NMR Structure Determination Jeffrey C. Hoch UConn Health Center Molecular, Microbial & Structural Biology Bioinformatics Fall 2004
2	Text for today’s lecture
3	EMBO practical course 2003: Structure determination of biological macromolecules by solution NMR
4	Structural constraints from NMR
5
6	How much information is needed to determine a protein structure? N atoms At least 3N–6 independent pieces of information (6 = 3 translational, 3 rotational degrees of freedom are meaningless) Assuming a rigid, well-defined structure
7
8	Over- or underdetermined? 100 residue protein ~ 1600 atoms ~4800 independent pieces of information 20-30 NMR restraints/residue Restraints highly correlated Structures not static Structure under-determined by NMR data alone
9	Additional structural constraints Sequence Prior knowledge of amino acid structure Covalent connectivity “Allowed” conformations S-S links
10	Protein dynamics The impact of dynamics on NMR parameters can be substantial Although the fluctuations are large, averaging usually makes measured parameters reflective of the average structure
11	Two main approaches Restrained molecular dynamics simulation Empirical potential energy function Hybrid distance geometry – simulated annealing Holonomic constraints (DG phase) Empirical potential energy function (SA phase)
12	MD simulation Thermal fluctuations of protein atoms can be treated classically using an empirical potential energy function U(x) Forces: Newton’s equations:
13	Empirical PE functions
14	Nonbonded interactions U_vdW and U_elec U_elecform depends on solvent model Distance-dependent dielectric implicit solvation: Implicit: N² interactions; reduce to ~30N via switching functions S Explicit: Periodic boundary conditions and Ewald summation
15
16
17
18
19
20
21
22	Structure Determination: Constrained Optimization Minimize E_emprical subject to the constraint that the structure is consistent with the experimental data Define E_experimental as a measure of the agreement between the structure and the experimental data
23	An equivalent unconstrained optimization
24	E_experimental
25	Existence and uniqueness: not guaranteed Neither E_empirical nor E_experimental are strictly convex --> multiple minima Many plausible structure (low E_empirical) are consistent (low E_experimental)
26
27
28
29	What are the appropriate weights for E_exp and E_empHow do we ensure different types of experimental data are treated with equanimity?
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45