Genetic Variation and Cancer an Industry Perspective

1
Genetic Variation and Cancer an Industry
Perspective

• Adrian Moody
• RD Genetics
• AstraZeneca

NCRI Genetic variation and Cancer workshop Friday
3rd February 2006
2
Overview

• Genetics in AstraZeneca
• Value of genetics data for cancer drug discovery
• Challenges of interpreting data
• Future perspective

3
(No Transcript)
4
Drug Discovery and Development Process
5
Value of Genetics Data to Cancer Drug Discovery

• Molecular characterisation and improved
  understanding of disease processes
• Opportunity to classify tumours by molecular
  signature rather than histology
• Target Identification through disease linkage
  (somatic germline)
• Personalised medicine opportunities to stratify
  patients for drug treatment with respect to
  response, adverse events, pharmacokinetics etc.
• Significant logistical / experimental advantages
  over other molecular measurements

6
Automated technology has shifted bottlenecks
AZs GLP-standard fully automated DNA archive and
reformatting system holds over 400,000 genetic
samples in carefully regulated conditions, stores
reformats gt5000 per day
High throughput genotyping facilities can achieve
gt50,000 reactions per day
With millions of datapoints generated within
weeks, data management and analysis has become
the new bottleneck in pharmacogenetic research
Automated DNA sequencers resequence gt1000
samples/ day
Thornton et al, Drug Discovery Today, 2005
7
Abundant public data

• Basic genetics data is increasingly plentiful
• Human Genome
• Db SNP
• HapMap
• Seattle SNP
• Cosmic
• Major new initiatives planned e.g. Cancer Genome
  Atlas Project at NCI

Location, Location, Location
8
An example - Gene Catalogue

• Allow genetic data to be referenced against the
  genome.
• Starting point for additional in silico
  analysis.
• Can be used to store additional relevant data
  i.e. population frequency etc.
• Limited use for providing broader context

9
Context
TATAGCTTGCATGGATG/TGACTC
10
Challenges for interpreting genetic data

• Distinguishing somatic mutations from germline
  polymorphisms in the absence of paired normal
• How many germline samples do you sequence to gain
  confidence in mutation data?
• Whats in a name?
• T790M EGFR mutation
• Integration of somatic and non-somatic genetics
• Somatic genetics and polymorphisms largely
  treated in isolation.

11
Challenges for interpreting genetic data

• Understanding functional consequence of genetic
  variation
• Sequence variation overlay against 3D structures
  if available, or primary sequence models which
  highlight key functional residues
• Copy number changes need to understand basis of
  amplification, amplicon boundaries and gene
  content
• Additional factors to understanding the
  importance of a mutation
• Causal, modifying or treatment selected?
• What tumour type, tumour stage, treatment regime?
• Additional genetic alterations

12
Challenges for interpreting genetic data

• Identifying relevant in vitro and in vivo models
  that recapitulate key clinical genetic variations
• Requires extensive characterisation of cancer
  cell lines
• Experimental validation of functional consequence
  required
• Cross community consistency in cell line and data
  standards
• Is your cell line the same as mine?
• Is my cell line the same now as it was 2 years
  ago?
• Genetics crosses disciplines and these data need
  to be integrated with biological, clinical and
  pharmacological phenotype.

13
Challenges for interpreting genetic data

• Statistical considerations for population based
  analysis.
• Prospective vs retrospective studies
• Is the phenotypic assessment consistent within
  the study and with other studies?
• Is the study suitably powered?
• Penetrance
• Ethnicity
• What to analyse?
• So much data, there must be something significant
  in here..

14
Future perspective

• Data analysis and integration recognised as a
  significant issue
• In-house initiatives
• Ensure valued added genetic data is accessible to
  all within the organisation.
• Integration of genetics with expression data and
  pharmacology
• Collaboration examples
• BBSRC industry interchange program proposal with
  Imperial Feasibility of multivariate methods for
  data integration
• Univariate/multivariate statistical methods to
  integrate different sources of biological data
  such as SNPs, gene expression, metabolic data and
  proteomics.
• Oxford High Dimension data, i.e. GWA

15

• Acknowledgements
• Tim French
• Thank you