Rebecca and John Moores UCSD Cancer Center Retreat

1
Toward a Digital Tumorigenesis Map of Breast
Cancer
Robert Stuart Cancer Biology William
Wachsman Viral Malignancy Chaitanya Baru San
Diego Supercomputer Center David
Tarin Translational Oncology
-Cancer Center Breast Carcinoma SPORE
component -SDSC NPEBC (Nat. Prog. Excellence in
Biomedical Computing)
2
Tumors complex societies
3
Reductionism
4
Crosstalk
5
Angiogenesis
Folkman, many others VEGF, bFGF, HGF, NOS VEGFR,
Ang1, Tie2
6
Inflammation and cancer
Hudson 1999 MIF from macrophages down regulates
p53 in fibroblasts Coussens 1999 Mast cells
infiltrate tumor edges and promote
angiogenesis Jacoby 1997 LOH at 10q22 in JPC
lamina propria immune cels
7
Stromal mesenchymal interactions
Olumi 1999 Ca-assoc fibroblasts direct prostate
tumor progression Many... -CAF produce ILGFm
HGF, FGFs, KGF, TGF-B. -CAF but not normal F
support various tumor cell types in
vitro -Genomic DNA abnormalities in breast CAF
8
Developmental Connection
e12
e13
e15
e17
nb
9
Successful Tumorigenic Cell will

• be a self starter w/r growth signals
• avoid apoptosis
• achieve limitless replicative potential
• find new markets (invasion/metastasis)
• co-opt the efforts of others into serving its own
  agenda
• angiogenesis
• growth signals from stroma
• etc

10
Difficult science

• Studies
• creation of appropriate stroma/premalignant in
  vitro models
• high resolution look at DNA
• high resolution view of RNA
• High resolution view of protein

• Development
• knockouts
• in vitro/growth factors
• in vitro/recombination

• Tumor
• cell lines biased toward independence

11
Digital Map of Tumorigenesis

• Detection of
• every transcript
• every protein including secreted
• every change in genomic DNA
• Single cell resolution
• Across time
• Comprehensive integration of domain knowledge

12
Visualization
13
Necessary components

• Small scale assay for class of interest
• Informatics infrastructure
• Database
• Statistical algorithms
• Visualization
• Data integration
• Reasonable short term goal

14
Small scale DNA array assay
15
Ng-scale assay in kidney development
16
SDSC local data architecture
LAN (multiple GbE, TCP/IP)
Local Disk (50TB)
Blue Horizon
WAN (40 Gb/s)
HPSS
Sun E10K
New Blue (4TF)
SAN (2 Gb/s, SCSI)
Via Cisco SCSI/IP
30 MB/s per drive
200 MB/s per controller
2GB/s
FC Disk Cache (150 TB)
FC GPFS Disk (50TB)
Database Engine
Data Miner
Vis Engine
Silos and Tape Drives (6 PB)
National Partnership for Advanced Computational
Infrastructure
San Diego Supercomputer Center
17
SKIDLkit J2EE Application and Web Service
SKIDL Client
Web Service
Service Registry UDDI / ebXML
RMI / IIOP
SOAP / WSDL
Application Server
EJB Container
SKIDL Client Session Bean
SKIDL Service Beans
Session Logs
Back-end Beans
SOAP / WSDL
JDBC
JNI
JNI
SRB Server
DB2, Oracle
Intelligent Miner
Web Service
National Partnership for Advanced Computational
Infrastructure
San Diego Supercomputer Center
18
A Realistic Starting Point

• Gene expression in stroma as a function of
  distance from malignant cells
• Microdissection and LCM
• Standard assay and ng
• Encode distance, cell types

19
GEP as f(cell type) in prostate biopsies
20
Long term goals/benefits

• Establish CC collaborations
• Integrate use of CC shared resources
• Microarray
• Histology
• Q-PCR
• Imaging
• Data compilation
• Strengthen SDSC links
• microarray informatics infrastructure
• leverage existing investment
• Preliminary data for outside funding!!!

21
Business 101 SWOT

• Strengths
• Comprehensive CC National SCC 1
• Experience in microarrays, ng-assays,
  mesenchymal-epithelial interactions, breast ca,
  data integration
• This is a Concept Proposal
• Weaknesses
• This is a Concept Proposal
• Opportunities
• Collaborations, SDSC, shared resources, funding
• Threats
• Someone else is going to do this, we ought to be
  first

U