1. Drug Discovery – Setting the Scene
Drug Discovery Challenges
Overview of course
Anne Hersey
ChEMBL Group, EMBL-EBI

2. Getting Drugs to Market is difficult and costly
Cost of bringing a new drug to market is $1.8billion
Clinical development (Phase 1-3) accounts for 63% of cost
Only 8% of new molecules make it from candidate selection to a marketed drug
Takes 13.5 years to discover and develop a drug
Need ~10 molecules a year entering clinical development to achieve 1 launched drug per year
Data from Steven Paul et al, NRDD (2010)

3. Discovery Development Med. Chem. SAR Drug Discovery Use ChEMBL content
Target Discovery
Lead Discovery
Lead Optimisation
Preclinical Development
Phase 1
Phase 2
Phase 3
Launch
(Phase 4)
Target identification
Microarray profiling
Target validation
Assay development
Biochemistry
Clinical/Animal disease models
Medicinal Chemistry
Structure-based drug design
Selectivity screens
ADMET screens
Cellular/Animal disease models
Pharmacokinetics
High-throughput Screening (HTS)
Fragment-based screening
Focused libraries
Screening collection
Toxicology
In vivo safety pharmacology
Formulation
Dose prediction
Safety
&
Efficacy
Indication
discovery, repurposing & expansion PK
tolerability
Efficacy
Discovery
Development
Use
Med. Chem. SAR
Clinical Candidates
Drugs
>1,600,000 compound records
>13,500,000 bioactivities
~60,000 abstracted papers
~10,000 targets
ChEMBL content
~12,000 clinical candidates
~1,600
drugs 3

4. Commonly asked questions:
I have a hit in a phenotypic screen
How can I identify what target(s) are involved
What target(s) are likely to affect my disease of interest?
What evidence is there for this?
Is this target druggable?
Is there a crystal structure for my target?
Are there any compounds known to bind to it?
I am interested in a compound/compound class that binds to my target
Which compounds are similar?
How potent and selective are these compounds?
What are their potential liabilities?
Have they been patented?
Can I buy these compounds?
Course can help answer these questions

5. How Many Biological Targets Are There?
Genes within the genome encode “target’ proteins
Bioactive molecules usually interact with proteins
Typical gene numbers in important genomes
Escherichia coli (a bacteria) ,377
Plasmodium falciparum (the malaria parasite) 5,268
Drosophila melanogaster (a fruit fly) ~17,000
Homo sapiens ~21,000 5

6. NFκB Pathway
Cell signalling, important in cancer, inflammation 6

7. FDA Approved Drugs 7

8. Clinical Candidates 8

9. How many chemicals are there?
GDB databases from Jean-Louis Reymond, University of Berne, Switzerland
GDB-13 database
Small organic molecules up to 13 atoms of C, N, O, S and Cl
Simple chemical stability and synthetic feasibility rules
977,468,314 structures
GDB-13 is the largest publicly available small organic molecule database 9

10. How Big is Bioactive Chemical Space?13
Heavy Atom Count
Likely to be ~1019 organic small molecules obeying Lipinski’s Rule of Five 10

11. Chemogenomics Exploration of bioactivity space at genomic scale
Structure Activity Relationship (SAR)
Drugs
Drug targets
102
Drugs
Screened proteins
103
Screened molecules
ChEMBL
All reasonable
molecules
Presented to P&G, Cincinnati, April 2005, © 2005 Inpharmatica Ltd.
All human
proteins
2x104 11

12. Chemical Space
Only certain molecules have features consistent with good pharmacological properties
All compounds
Available compounds
Drug-like compounds 12

13. Target Space
Only certain targets have binding sites capable of ligand efficient binding of drug-like ligands
All targets
Available targets
Druggable targets 13

14. Accessible Pharmacological Space
Drug-like compounds but no complementary targets
Druggable targets and complementary compounds
Available compounds for target but non-drug-like
Druggable targets but no complementary compounds 14

15. Also ... Drugs need to be safe & have good systemic exposure
Drug-like compounds but no complementary targets
Available compounds for target but non-drug-like
Druggable targets and complementary compounds and safety and exposure
“Safety Space”
“Exposure Space” 15

16. Known/similar target structure Unknown target structure
Project Idea
Phenotypic assay
Target-based assay
Novel
Previously screened
Known/similar target structure
Unknown target structure
Known/similar ligands
In-house compound collection
Commercial compounds
Virtual compounds
HTS Screen
Purchase of specific compounds
Commission synthesis
Drug 16

17. Computational Tools

18. Where to start? Help is at hand ....

19. Who is working on drug discovery?
Large Pharma Companies
Small – Medium Enterprises (SMEs)
Academia
Not for Profit organisations
Perspectives from:
Large Pharma (Darren Green, GSK)
Not for Profit (Benoit Laleu, MMV)
Public/Private Partnerships (Ian Dunham, EMBL-EBI (CTTV)) 19

20. Choosing Targets Using GWAS Studies (Aroon Hingorani, UCL)
Druggable Genome (Anna Gaulton, EMBL-EBI)
Bioinformatics and canSAR (Bissan Al-Lazikani, ICR)
Systems Biology (Julio Saez-Rodriguez, Aachen University Hospital)
Using 3D protein Structures (John Berrisford, EMBL-EBI)
CTTV (Centre for Target Validation) 20

21. Choosing Compounds Bioactivity Databases (Anne Hersey, EMBL-EBI)
Purchasable Compounds (John Irwin, UCSF)
Aggregators (John Irwin, UCSF)
Ligand and Structure-based design (Val Gillet, University of Sheffield)
Chemoinformatics (Nathan Brown, ICR)
Patent Databases (George Papadatos, EMBL-EBI)
Use Cases (Darren Green, GSK) 21

22. Phenotypic Screens to Targets
Target identification (Grace Mugumbate, EMBL-EBI)
SEA (John Irwin, UCSF)
Drug Repurposing
Strategies and practical (John Overington, Stratified Medical) 22

23. Tools, Software and Work Flows
Open Source Chemoinformatics Software (Nathan Brown, ICR)
myChEMBL - ipython notebooks, RDKit (Michal Nowotka, EMBL-EBI)
Work flow tools - Knime (George Papadatos, EMBL-EBI)
Data Visualisation – (Nathan Brown, ICR)
Cross referencing chemical strcutures – UniChem (Jon Chambers, EMBL-EBI) 23

24. Summary Selecting Targets Selecting Compounds
Using Computational Tools to do the above
Ask questions, discuss etc
Enjoy the course! 24