David Ryan Koes

Structure-based search of chemical libraries with

PharmitDavid Ryan Koes

Computational and Systems Biology University of Pittsburgh

@david_koes

2http://pharmit.csb.pitt.edu/

3

24 x 10 TB HDD 4 TB SSD Cache

512 GB RAM 40 cores

4

5

Server

Client

6

http://pharmit.sf.net GPL/BSD License

ShapeDB PharmerOpenBabel

RDKit

3Dmol.jsJQuery

pharmitserver

OpenBabel

pharmit.js

smina

ChemDiv

MolPortNCI

PubChem

Prebuilt Libraries

User LibrariesSearch

Minimization OpenBabel

buildlibs.py

Chemspace CHEMBL

ZINCMCULE

Client

3Dmol.jsJQuery

pharmit.js

Why web application?

8

http://3dmol.csb.pitt.edu Rego, N., & Koes, D. (2014). 3Dmol.js: molecular visualization with WebGL.

Bioinformatics. doi:10.1093/bioinformatics/btu829

Aside: Molecular Active Learning

9

Demo

Go to this URL: http://3dmol.csb.pitt.edu/viewer.html

Libraries

10

ChemDiv

MolPortNCI

PubChem

Chemspace CHEMBL

ZINCMCULE

pharmit currently has8 built-in libraries containing 1,183,898,546

conformations of 208,783,424 compounds and 30 publicly

accessible user-contributed libraries containing

18,932,611 conformations of 1,588,877 compounds.

86% New Compounds

Library Usage

12

Contributed Libraries

13

14

pharmitserver

ShapeDB smina

http://pharmit.sf.net GPL/BSD License

Pharmer

15

HydrophobicFeatures

Hydrogen Acceptor Feature

Hydrogen Donor

Feature

Hydrogen Donor

Feature

Pharmacophore A spatial arrangement of molecular features essential for biological activity

Pharmer Efficient and Exact Pharmacophore Search

Koes, D. R., & Camacho, C. J. (2011). Pharmer: efficient and exact pharmacophore search. Journal of Chemical Information and Modeling, 51(6), 1307-1314. doi:10.1021/ci200097mKoes, D. R., & Camacho, C. J. (2012). ZINCPharmer: pharmacophore search of the ZINC database. Nucleic acids research, 40(Web Server issue), W409-414. doi:10.1093/nar/gks378

16

HydrophobicFeatures

Hydrogen Acceptor Feature

Hydrogen Donor

Feature

Hydrogen Donor

Feature

Pharmer Efficient and Exact Pharmacophore Search

17

Pharmer Efficient and Exact Pharmacophore Search

18

q

a b

cd

ef

g

h

split xa - h

x

ysplit ya - d

split ye - h

split xa - b

split yc - d

split xe - f

split yg - h

cb d hgfea

q

a b

cd

ef

g

h

split xa - h

x

ysplit ya - d

split ye - h

split xa - b

split yc - d

split xe - f

split yg - h

cb d hgfea

Pharmer Efficient and Exact Pharmacophore Search

q

a b

cd

ef

g

h

split xa - h

x

ysplit ya - d

split ye - h

split xa - b

split yc - d

split xe - f

split yg - h

cb d hgfea

19

Pharmer Efficient and Exact Pharmacophore Search

20

Pharmer Efficient and Exact Pharmacophore Search

0.1!

1!

10!

100!

1000!

0.1! 0.2! 0.3! 0.4! 0.5! 0.6! 0.7! 0.8! 0.9! 1!

Tim

e (s

)!

Tolerance Sphere Radii Multiplier!

MOE HSP90!MOE FXIa!Pharmer HSP90!Pharmer FXIa!

21

pharmitserver

smina

http://pharmit.sf.net GPL/BSD License

Pharmer ShapeDB

22

ShapeDB Indexed Search of Molecular Shapes

Align to Moments of Inertia Voxelize

Oct-tree • Scales with Surface Area, not Volume • Fast Intersection/Union Operations

23

ShapeDB Indexed Search of Molecular Shapes

0.01

0.1

1

10

100

1000

10^4 10^5 10^6 10^7

Aver

age

Tim

e (s

)

Magnitude of Search Space

Performance of Shape Similarity Search

Similarity (Scan)

24

MIV MSVMIVMSV

MSV Minimum Surrounding Volume

Union

MIV Maximum Included Volume

Intersection

Matching and packing algorithm for efficient and effective initialization

ShapeDB Indexed Search of Molecular Shapes

Koes, D. R., & Camacho, C. J. (2014). Shape-based virtual screening with volumetric aligned molecular shapes.J Comput Chem, 35(25), 1824-1834. doi:10.1002/jcc.23690Koes, D., & Camacho, C. (2014). Indexing volumetric shapes with matching and packing. Knowledge and Information Systems, 1-24. doi:10.1007/s10115-014-0729-z

25

ShapeDB Indexed Search of Molecular Shapes

Shape Constraints

26

ShapeDB Indexed Search of Molecular Shapes

0.0001

0.001

0.01

0.1

1

10

100

1000

10^4 10^5 10^6 10^7

Aver

age

Tim

e (s

)

Magnitude of Search Space

Performance of Shape Constraint Search

Search Constraints (Indexed) Similarity (Scan)

27

pharmitserver

ShapeDB smina

http://pharmit.sf.net GPL/BSD License

Pharmer

28

smina Scoring and Minimization with AutoDock Vina

Improved energy minimization

Facilities for scoring function development

Custom format with precomputed torsion tree

Implementation of server protocol

Koes, D. R., Baumgartner, M. P., & Camacho, C. J. (2013). Lessons Learned in Empirical Scoring with smina from the CSAR 2011 Benchmarking Exercise. Journal of Chemical Information and Modeling, 53(8), 1893-1904. doi:10.1021/ci300604z

Case Study: Profilin

29

30

Top 20 compounds purchased

31

Acknowledgements

32

http://pharmit.csb.pitt.edu

Department of Computational and Systems Biology

NIGMS R01GM108340

Google Cloudgithub.com/gnina

http://bits.csb.pitt.edu

@david_koes

33

Questions?