Overview & Applications. T. Lezon Hands-on Workshop in Computational Biophysics Pittsburgh Supercomputing Center 04 June, 2015

Transcription

1 Overview & Applications T. Lezon Hands-on Workshop in Computational Biophysics Pittsburgh Supercomputing Center 4 June, 215

2 Simulations still take time Bakan et al. Bioinformatics 211. Coarse-grained Elastic Network Models are fast Lane et al

3 The structural data explosion Multiple structures for a single sequence Nature, 15 May 214. Dynamics may be inferred from structural data. 3

4 Exploiting the PDB since 21 High-throughput analysis of structural data Application Programming Interface (API) for development of tools Suitable for interactive usage p38 ensemble (PCA) Experiment/Theory User inputs a sequence Usage example >1A9U:A PDBID CHAIN GSSHHHHHHSSGLVPRGSHMSQ ERPTFYRQELNKTIWEVPERYQ NLSPVGSGAYGSVCAAFDTKTG... identifies, retrieves, aligns, and analyzes (PCA) structures matching input sequence Bakan, Meireles & Bahar. Bioinformatics 211. p38 network model (ANM) MD trajectory analysis (EDA) User can Compare experimental and theoretical models Sample conformations along normal modes 4

5 An Interactive Tool 5

6 An evolving suite of tools Principal Component Analysis Elastic Network Models Normal Mode Analysis Trajectory Analysis Multiple Sequence Alignment Correlated Mutation Analysis Structural Evolution Computational Drug Discovery Binding Site Prediction Affinity Estimation Call ProDy from VMD Normal Mode Visualization 6

7 Tutorials: ProDy & Structure Analysis Obtaining PDB Files BLAST Searching the PDB Constructing Biomolecules from Transformations Aligning and Comparing Structures Identifying Intermolecular Contacts 7

8 Tutorial: Elastic Network Models Gaussian Network Model (GNM) Anisotropic Network Model (ANM) Normal Mode Algebra Deformation Analysis Customizing ENMs 8

9 Elastic Network Model Useful for finding global equilibrium motions of proteins Employs harmonic potential about native state Coarse-grained (Cα-only description) Residue pairs are connected via springs Normal modes are found analytically 9

10 Elastic Network Model Useful for finding global equilibrium motions of proteins Employs harmonic potential about native state Coarse-grained (Cα-only description) Residue pairs are connected via springs Normal modes are found analytically 1

11 Elastic Network Model Useful for finding global equilibrium motions of proteins Employs harmonic potential about native state Coarse-grained (Cα-only description) Residue pairs are connected via springs Normal modes are found analytically 11

12 Flexible force constants Tirion, PRL 77 (1996). Hinsen et al. Proteins 33 (1998). Hinsen et al. Chem Phys 261 (2). Yang et al. PNAS 16 (29). 12

13 Optimizing force constants Download NMR structures from PDB Calculate residue MSFs for each protein Assign ENM topology fetchpdb() calcmsf() Optimize force constants to reproduce structural dynamics Search for trends in force constant values with structure buildhessian() 13

14 Fine-tuning force constants R=.91 R=.91 Distance-dependence 1 st neighbors 2 nd neighbors H bonds R=.76 R=.79 black: NMR red: ENM blue: modified ENM R=.31 R=.57 GammaStructureBased() Learn more at prody.csb.pitt.edu 15

15 Tutorial: Ensemble Analysis NMR Models Homologous Proteins Multiple X-ray Structures Multimeric Proteins 16

16 Example: Comparing PCA and ENM Structures of HIV1-RT Unbound Inhibitor bound DNA bound Bakan & Bahar. PNAS 16 (29). 17

17 Example: Comparing PCA and ENM Structures of p38 MAPK Unbound Inhibitor bound Glucose bound Peptide bound Bakan & Bahar. PNAS 16 (29). 18

18 Tutorial: Trajectory Analysis Fast processing of long trajectories Enables comparison of MD trajectories and structural data or ENM results 19

19 Tutorial: Conformational Sampling Sample structures along normal modes Refine structures using NAMD 2

20 Tutorial: Druggability Set up NAMD simulations Analyze trajectories to identify binding hot spots 21

21 Exploring binding with probe molecules Bakan et al. J Chem Theor Comput (212). 22

22 Tutorial: Evol 23

23 Tutorial: Normal Mode Wizard 24

24 Global transitions Outward Facing (OF) Inward Facing (IF) 3Å Yernool et al. Nature 431 (24). Reyes et al. Nature 462 (29). 25

25 Global transitions Single subunit showing the transport domain moving across the membrane Reyes et al. Nature 462 (29). 27

26 Rotations-Translations of Blocks H RTB P (6N b 3N) (6N b 6N b ) Smaller Hessian can be more easily diagonalized... H AA (3N 3N) P T (3N 6N b ) V RTB V AA P T...and modes projected back into all-residue space H: ANM Hessian (3 rows/cols per residue) P: Projection matrix from all-residue space to rigid block space H RTB : RTB Hessian (no internal motions of blocks) V AA : Approximate ANM motions RTB.buildHessian() Ming & Wall. PRL 95 (25). Zheng & Brooks. Biophys J 89 (25). 28

27 Exploring structural transitions: Glutamate transporter ANM predicts large radial motions of the trimer. Can we invent a better model? H R 2 x x y x z 2 x y y y z 2 x z y z 2 z H Lezon & Bahar. Biophys J 12 (212). H = - Altered radial force constants: R g 2 ( R ) 2 x x y x z é ê ê ê ê ê ëê x x x 2 y z x y ( x ) 2 x y x y cx z y y z y x y 2 cx z ( y ) 2 cy z cy z ( cz ) 2 y z ù ú ú ú ú ú ûú x y z z 2 z z x y z z RTB.buildHessian() 3

28 Exploring structural transitions: Glutamate transporter ANM: Large radial motions imanm 32

29 Explicit membrane models system H ss H ES H SE H EE environment As the environment fluctuates randomly, the effective motion of the system is given by reducemodel() 33