Transcript Dr. Sharmila Mande

Genomes to Drugs:
A Bioinformatics Perspective
Sharmila Mande
Bioinformatics Division
Advanced Technology Centre
Asia’s Largest Global Software & Services Company
Data types along the drug
discover chain
Target
Identification
Target
Lead
Validation
Identification
Genomics
SNP
EST/cDNA
Lead
Optimisation
Chemistry
ADMET
HTS
SNP
Pre-clinical
trials
Gene expression
Proteomics
Asia’s Largest Global Software & Services Company
Traditional drug discovery process
• Development time per drug: 8 to 12 years.
• Development cost per drug: $500- $900 M
• High failure rates - especially in clinical stage
Asia’s Largest Global Software & Services Company
Post-genomics scenario
• Reduction of drug development time
• Reduction of cost of developing drugs
– Develop high quality information early in drug
discovery process
– Faster identification of new drug targets
– Higher quality target and leads
– Reduction of drug failure rates
Asia’s Largest Global Software & Services Company
Design principles
• Do experiments in silico first
Hypothesis
Selection
Synthesis
“Real World”
Bioassay
Analysis
“in silico World”
Asia’s Largest Global Software & Services Company
Whole genomes
• Complete genome sequences of
– about 138 microbial organisms
– Human
– Parasites
– Plant
• Possible to develop novel cures
Asia’s Largest Global Software & Services Company
Drug Target Identification
Three possible cases:
• Human protein is not functioning properly, or its
activity needs to be modified.
Sickle cell anemia
• The potential target is a key protein from infectious
organism, and has no counterpart in humans
Bacterial Cell wall
• The target is a protein from an infectious organism
and a homologous protein exists in humans.
DHFR
Asia’s Largest Global Software & Services Company
How Many Drug Targets ?
• Present day therapy addresses only 500 molecular
targets
• Cell membrane receptors constitute the largest
class of current drug targets
• Human and other microbial genomes suggest that
many more targets may be available
Asia’s Largest Global Software & Services Company
Target Identification: Comparative
Genomics
• Target identification in the pre-genomics era was an
extremely tedious and time consuming process.
• Genomics approaches - a major development in
recent times in drug discovery
e.g. broad spectrum antibiotics
drugs against malaria parasite
Asia’s Largest Global Software & Services Company
Target Validation
• There is an opportunity to identify many more
targets
• Bottleneck in validation of drug targets
Asia’s Largest Global Software & Services Company
Computational Approaches in Drug
Design
without prior knowledge of
receptor structure
with prior knowledge of
receptor structure
Generation of a
pharmacophore
3D arrangement of
functionally important parts
of molecules
Docking
Scan a database for optimal
fitting of inhibitors
Quantitative structureactivity relationships
De-novo design
Generate a “custom” ligand
for a given active site
Asia’s Largest Global Software & Services Company
Drug Design Cycle
Asia’s Largest Global Software & Services Company
Computational Approaches in Drug Design:
No prior knowledge of receptor structure
Combinatorial Chemistry
Asia’s Largest Global Software & Services Company
Pharmacophore
generation
Cyclic urea inhibitor
of HIV protease from
pharmacophore
hypothesis.
Asia’s Largest Global Software & Services Company
Computational Approaches in Drug Design:
Prior knowledge of receptor structure
• Captopril, the first therapeutic
drug from structure based
studies
• Antihypertensive
• Angiotensin converting
enzyme modelled on
carboxypeptidase
• 30,000 fold improvement in
inhibitory activity obtained,
from the first lead Nsuccinoyl-prolin to captopril
Asia’s Largest Global Software & Services Company
Examples of Structure Based Drug
Design
HIV protease with amprenavir
Influenza with neuraminidase
inhibitor
Asia’s Largest Global Software & Services Company
If HTS is available, why do docking at all?
Protein Tyrosine Phosphatase 1B inhibitors:
• 400,000 compounds screened by HTS vs. 365
compounds high scoring docking compounds.
• Hit-rate by computational approaches 1700 times
higher. Computationally identified compounds more
“drug-like”.
Dihydropicolinate reductase inhibitor:
• hit rate < 0.2% in HTS and > 6% in computer-based
approach
Asia’s Largest Global Software & Services Company
Lead Optimization
active site
Receptor
/ Enzyme
Enzyme
Receptor
/ Enzyme
Enzyme
Ki~1mM
Receptor/
Enzyme
Enzyme
Ki~1-10 nM
Asia’s Largest Global Software & Services Company
ADMET
Drug failures
• 39% Poor pharmacokinetics
• 11% Toxicity
ADMET prediction
• in-silico approaches
Asia’s Largest Global Software & Services Company
In silico ADMET
Identify
• Non soluble compounds
• Non permeable compounds
• Non metabolically stable compounds
• Toxicophores
Asia’s Largest Global Software & Services Company
Bioinformatics success stories
• Quick target identification (Merck)
– potential drug targets for schizophrenia
• Expedited drug candidate identification (SKB)
– drugs for bone tumor (target cathepsin K)
• Poor drug candidate elimination (Aventis vs GSK
and SP)
– anti IL-5 therapy against asthma
Asia’s Largest Global Software & Services Company
TCS’ Bioinformatics Division
• Part of Advanced
Technology Centre
Genome
Analysis
Sequence
Analysis
Comparative
Genomics
• 40 persons (9 Ph.D’s)
TCS
Drug Design
• Undertaking R&D in
selected areas of
computational biology
Bio-Suite
Structural
Analysis
3D
Modelling
Simulation
Structural
Manipulations
Asia’s Largest Global Software & Services Company
Thank You
Asia’s Largest Global Software & Services Company