Using AutoDock With AutoDockTools: A Tutorial

1
UsingAutoDock WithAutoDockToolsA Tutorial

• Ruth Huey
• Garrett M. Morris
• Michel F. Sanner

2
Docking in the Scheme of Things

• world health
• drug discovery
• identification of target
• small molecule therapeutic agent
• gt95 focused on protein binding
• computational chemistry
• docking programs
• Dock, AutoDock,
• FlexX, Glide, Gold, .

4 drugs bound to HIV protease
3
What is Docking?

• Best ways to put two molecules together.
• Three steps
• (1) Definition of the structure of the target
  molecule.
• (2) Location of the binding site.
• (3) Determination of the binding mode.

4
What is Docking?

• Best ways to put two molecules together.
• Need to quantify or rank solutions
• Scoring function or force field.
• Best ways to put two molecules together.
• (plural) Experimental structure may be amongst
  one of several predicted solutions.
• Best ways to put two molecules together.
• Need a Search method

5
AutoDock Introduction

• Automated docking of flexible ligands to
  proteins.
• Global search algorithms
• Simulated Annealing (Goodsell et al. 1990)
• Distributed SA (Morris et al. 1996)
• Genetic Algorithm (Morris et al. 1998)
• Local search algorithm
• Solis Wets (Morris et al. 1998)
• Hybrid global-local search algorithm
• Lamarckian GA (Morris et al. 1998)
• Empirical free energy function estimates Ki
  (Std. dev. 2 Kcal mol-1)

6
AutoDock 3Scoring Function

• DGbinding DGvdW DGelec DGhbond
• DGdesolv DGtors
• DGvdW
• 12-6 Lennard-Jones potential
• DGelec
• Coulombic with Solmajer-dielectric
• DGhbond
• 12-10 Potential with Goodford Directionality
• DGdesolv
• Stouten Pairwise Atomic Solvation Parameters
• DGtors
• Number of rotatable bonds

7
AutoGrid Why Use Grid Maps?

• AutoGrid computes grid maps
• Representation of macromolecule
• Regular orthogonal lattice of points
• Ligand probe samples force field
• One map for each ligand atom type
• AutoDock uses trilinear interpolation
• to compute interaction energy between
• ligand and target
• Non-bonded energy is pre-calculated
• Saves time 100x faster

8
AutoDock Citations
3424 AutoDock Licenses as of 9/8/05
Source ISI Science Citation Index
9
python arrives at mgl

• Michel Sanner (1997 IPython San Jose)
• DejaVu OpenGL-based viewer
• PMV PythonMolecularViewer
• independent, reusable-component based
• multiple molecules, multiple representations
• interactive picking and access to python
  interpreter
• www.scripps.edu/sanner

10
Python-based Tools for AutoDock

• Graphical User Interface
• Scriptable Python Objects
• AutoDock Scorers

11
what python has done for us part 1

• Graphical User Interface
• prepare input files
• define 3D search space
• analyze results

12
ADT AutoDockTools

• Application specific extension of PMV
• Set of AutoDock specific PMV commands
• ADT programmer does not need to be a specialist
  in OpenGL, PDB parsing, etc.
• Provides all of PMVs functionality
• High level of code-reuse
• Example of extending and customizing PMV to
  facilitate automated docking

13
ADT Input Preparation Tools
14
ADT Docking Analysis Tools
15
ADT Tutorial

• 11 hands-on exercises
• prepare 4 input files
• ligand, receptor
• parameter file for autogrid
• parameter file for autodock
• launch autogrid
• launch autodock
• analyze results
• 25 sessions, gt250 people
• Handout pdf input files available
  www.scripps.edu/mb/olson/doc/autodock

http//w3.to/autodock
16
what python has done for us, part 2

• Scriptable Python Objects
• model-view-controller
• MoleculePreparation
• GridParameters, DockingParameters
• Docking, Cluster
• object-oriented API
• virtual screening

17
Input File Preparation
18
Using AutoDock for Virtual Screening

• William Lindstrom, Garrett Morris, Christoph
  Weber, Ruth Huey

19
Virtual Screening

• Use of high-performance computing to analyze
  large databases of chemical compounds in order to
  identify possible drug candidates W.P. Walters,
  M.T. Stahl and M.A. Murcko, Virtual Screening-An
  Overview, Drug Discovery Today, 3, 160-178
  (1998). REOS (rapid elimination of swill)

20
NCI Diversity Set

• Almost 140,000 compounds in NCI repository
• 71,756 compounds, for which gt1.0 gram available
• Used Chem-X (Oxford Molecular Group) to generate
  3-point pharmacophores to create a set of
  1,000,000 pharmacophores for all acceptable
  conformations of each structure. Then used Chem-X
  to build a diverse subset by comparing all
  pharmacophores for each acceptable conformation,
  adding the structure to the set if it had 5 or
  more new pharmacophores.
• 1990 compounds, with diversity based on unique
  3-point pharmacophores
• A unique set of 1990 input ligand files for
  AutoDock available from cvs repository rotatable
  bonds defined, atomic charges added and inspected
  for quality

21
e.g. AICAR transformylase

• AutoDock 3 was used to screen the NCI Diversity
  Set, 1990 compounds
• 5 million evals per run
• 100 runs per compound
• 2 weeks on 32 nodes of Linux cluster

22
AICAR Results of virtual screening and kinetic
inhibition testing

• In silico
• 44 top compounds, Ebinding lt -13.0 Kcal/mol
• In vitro
• 10 are insoluble in water
• 18 precipitate in buffer solution
• 8 out of 16 soluble compounds bind
• (50 success)
• Li, Chenlong et al. J. Med. Chem. Dec. 2004,
  47(27), 6681-6690

23
what python has done for us, part 3

• Scorers based on AutoDock3 Force Field
• python prototype scorers
• c scorers
• Representation of Molecules
• MolKit MolecularSystem

24
AutoDock Scorers
25
Using AutoDock Scorers

• Energy-per-term for recalibration
• intermolecular
• internal
• FlexTree, work by Yong Zhao and Michel Sanner

26
FlexTree-based ComputationsAutomated
DockingYong Zhao and Michel Sanner
27
Protein Flexibility
Flexibility Tree
Coarsness
Hinge
28
Acknowledgements

• Michel Sanner
• Garrett Morris
• David Goodsell
• William Lindstrom
• Yong Zhao
• Peggy Graber
• Alex Gillet
• Art Olson
• NIH R01 GM069832, NIH P41 RR08605