INTERNATIONAL COLLABORATION IN PROTEOMICS AND INFORMATICS

1
INTERNATIONAL COLLABORATION IN PROTEOMICS AND
INFORMATICS

• Bibliotheca Alexandrina, 9 October, 2007
• Gilbert S. Omenn, M.D., Ph.D.
• Center for Computational Medicine Biology
• Chair, HUPO Plasma Proteome Project
• University of Michigan, Ann Arbor, MI, USA

2
It Is Such A Great Pleasure to Visit The
Bibliotheca Alexandrina

• One of the Wonders of the Modern World!
• The First Digital Library, from its Birth
• Facilitating International Collaboration in
• Science and Technology

3
Nearly-Complete Human Genome Sequence, 15-16 Feb
2001
4
We Live in a New World of Life Sciences

• New Biology---New Technology a parts list
• Genome Expression Microarrays
• Comparative Genomics CNV miRNA
• Proteomics and Metabolomics
• Bioinformatics Computational Biology

• Mechanism- Evidence-Based Medicine
• What were you doing up to now?!
• Predictive, personalized, preventive,
• participatory healthcare and community
• health services

5
Key Components of the Vision of Biology As An
Information Science

• An avalanche of genomic information validated
  SNPs, haplotype blocks, candidate genes/alleles,
  proteins, metabolites--associated with disease
  risk
• Powerful computational methods
• Effective linkages with better environmental and
  behavioral datasets for eco-genetic analyses
• Credible privacy and confidentiality protections
• Breakthrough tests, vaccines, drugs, behaviors,
  and regulatory actions to reduce health risks and
  cost-effectively treat patients globally.

6
A Golden Age for the Public Health Sciences

• Sequencing and analyzing the human genome is
  generating genetic information that must be
  linked with information about
• Nutrition and metabolism
• Lifestyle behaviors
• Diseases and medications
• Microbial, chemical, physical exposures
• Every discipline of public health sciences
  needed.

7
Definitions

• Genetics is the scientific study of genes and
  their roles in health and disease, physiology,
  and evolution.
• Genomics is a modern subset of the broader field
  of genetics, made feasible by remarkable advances
  in molecular biology, biotechnology, and
  computational sciences, to examine the entire
  complement of genes and their actions.
• Global analyses permit us and require us to go
  beyond the known lamp-posts of individual gene
  associations and effects.

8

• Proteins are the action molecules of the cell and
  the leading candidates for biomarkersin tissues
  and in the blood. Proteins are coded for by
  genes. Understanding one protein can be a
  lifetimes work!
• Proteomics is the global analysis of proteins in
  cells or body fluids. Techniques for global
  analysis of proteins are advancing rapidly,
  especially for discovery of biomarkers for
  diagnosis, treatment, and prevention.
• Metabolomics is the global analysis of
  metabolites.
• Proteomics metabolomics epigenomics
  functional genomics

9
Protein
DNA
10
Rationale for Proteomics

• Proteins are much closer to the pathophysiologic
  changes and molecular targets for drugs than are
  mRNAs.
• Changes in mRNAs are clues, but changes in
  corresponding proteins often are not highly
  correlated.
• Advances in fractionation of complex tissue and
  plasma protein mixtures, in mass spectrometry,
  and in curated databases of proteins help address
  complexity, dynamic range, and uncertainty of
  protein identifications.

11
A Vision For Proteomics

• Multiple protein biomarkers discovered
• Biomarkers combined on diagnostic chips
• Detect organ location of cancers, for surgery or
  radiation
• Detect mechanism of disease for chemotherapy,
  even if location unknown
• Mechanistic, rather than geographic
  classification
• Better efficacy/less toxicity for all types of
  patients

12
Status of Proteomics Assays

• Many technology platforms of increasing
  sensitivity and resolution
• Patterns or specific proteins still just
  biomarker candidates most lack independent
  confirmation and coefficient of variation, let
  alone validation with standard clinical
  chemistry parameters of sensitivity, specificity,
  and especially positive predictive value
• Approaches of clinical chemistry needed to guide
  further development of the field

13
Barriers for Proteomic Cancer Biomarker Discovery
in Plasma

• Human cancers are very heterogeneous
• Tumor proteins are in low abundance for early
  detection of cancers
• Tumor proteins are greatly diluted upon release
  to ECF and blood
• Plasma is an extraordinarily complex specimen
  dominated by high abundance proteins (50 by
  weight is albumin)
• Knowledge of the plasma proteome is still limited

14
Outline of Lecture

• Review of the vision, strategy, and output of the
  HUPO Human Plasma Proteome Project Pilot Phase
• Objectives for the New Phase of the Plasma
  Proteome Project
• Example of the power of computational tools and
  collaborations (if time)

15
HUPO

• The international Human Proteome Organization
  (HUPO) was founded in 2001. Its aims are
• To advance the science of proteomics
• To enhance training in proteomics
• To build international initiatives by organ
  (liver, brain, kidney), biofluid (plasma, urine,
  CSF, saliva), and disease (cardiovascular,
  cancers), plus antibodies and data standards.

16
Proteomics Interaction Map Ruth McNally,
sociologist
17
Samir Hanash, founding President of HUPO
Gil Omenn, leader of HUPO PPP
18
THE PLASMA PROTEOME

• Advantages The most available human specimen
  the most comprehensive sample of tissue-derived
  proteins the basis for a Disease Biomarkers
  Initiative tied to organ proteomes.
• Specific Disadvantages
• Extreme complexity/enormous dynamic range
• High risk of ex vivo modifications
• Lack of highly standardized protocols
• General Challenges Inadequate appreciation of
  incomplete sampling by MS/MS evolving
  annotations and unstable databases

19

• Long-Term Scientific Goals of the HUPO
• Human Plasma Proteome Project
• 1. Comprehensive analysis of plasma and serum
• protein constituents in people
• Identification of biological sources of variation
  
• within individuals over time, with
  validation of
• biomarkers
• Physiological age, sex/menstrual cycle,
  exercise
• Pathological selected diseases/special
  cohorts
• Pharmacological common medications
• 3. Determination of the extent of variation
  across
• populations and within populations

20
Scheme Showing Aims and Linkages of the
HUPO Plasma Proteome Project, Pilot
Phase
Serum vs Plasma
Technology Platforms--Separation and
Identification
Reference Specimens
HUPO HUMAN PLASMA PROTEOME PROJECT (PPP)
Development Validation of Biomarkers
HUPO PPP Participating Labs
Technology Vendors
Liver and Brain Proteome, Antibody, Protein Stds
Projects
Omenn GS. The Human Proteome Organization Plasma
Proteome Project Pilot Phase Reference
Specimens, Technology Platform Comparisons, and
Standardized Data Submissions and Analyses.
Proteomics 200441235-1240.
21
OUTPUT FROM PPP Pilot Phase

• Special Issue Aug 2005, Proteomics, Exploring
  the Human Plasma Proteome 28 paperscollaborativ
  e analyses and annotations, plus lab-specific
  analyses, and Wiley book (2006)
• Publicly-accessible datasets
• www.ebi.ac.uk/pride EBI www.peptideatlas.org
  /repository ISB
• www.bioinformatics.med.umich.edu/hupo/ppp
• Additional papers are encouraged
• Nature Biotechnology 2006 24333-338
  (States et al)
• Genome Biology 20067R35 (Fermin et al)
• Proteomics 2006 6 5662-5673 (Omenn)
• Numerous citations/comparisons of datasets

22
(No Transcript)
23
SERUM AND PLASMA REFERENCE SPECIMENS

• BD specially prepared male/female pooled
  samples, divided into EDTA-, Heparin-, and
  Citrate-anti-coagulated Plasma and Serum (250 ul
  x4 of each).
• BD clot activator. No protease inhibitors.
  Three separate ethnic pools prepared. Shipped
  frozen.
• 2. Chinese Academy of Medical Sciences Sets of
  three
• plasmas serum, similar to BD protocol.
• 3. National Institute for Biological Standards
  Control,
• UK citrate-anti-coagulated, freeze-dried
  plasma, from
• 25 donors, prepared for Intl Soc Thrombosis
  
• Hemostasis, 1 ml aliquots/ampoules.

24
Specifications for Data Submission

• Each of 55 labs agreed (July, 2003 Workshop) to
  provide, and 31 labs did provide
• a) a detailed experimental protocol, to push
  the limits to detect low-abundance proteins
• b) peptide sequences, rated as high or lower
  confidence, based on MS/MS criteria
• c) protein IDs from IPI 2.21 (July 2003) and
  search engine parameters used to align peptide
  sequences with proteins in human database
• Later, we obtained m/z peak lists and raw spectra
  (by DVD) for independent analyses.

25
From Peptides to Genome Annotation
digestion
databasesearch
LC-MS/MS
extraction
Peptides
Mass Spectrum
Proteins
Sample
Peptides
Spectrum Peptide Probability
Spectrum 1 LGEYGH 1.0
Spectrum N EIQKKF
0.3
statistical filtering
BLAST protein database
SBEAMS
Map to genome
Peptide Chrom Start_Coord
End_Coord PAp00007336 X
132217318 132217368


visualization
PeptideAtlas Database
Genome Browser
26
Numbers of Proteins Identified (LC-MS/MS or
FTICR-MS, 18 labs)

• From 15,519 reported distinct protein IDs in IPI
  2.21, we chose one representative/cluster
• (a) 9504 1 or more peptide matches
• (b) 3020 2 peptide matches (Core Dataset)
• (c) 1274 3 or more peptide matches
• 889 follow-up high-stringency analysis with
  adjustments for protein length and multiple
  (43,000) comparisons in IPI v2.21
• (Nature Biotech 2006 24333-338)

27
GREATEST RESOLUTION AND SENSITIVITY

• The most extensive high-confidence yield was from
  combined methods of immunoaffinity (top-6)
  depletion, 2 or 3-D high-resolution
  fractionation, and then ESI-MS/MS with ion-trap
  LTQ instrument.
• LTQ gave several fold more IDs (1168) than did
  LCQ (271) in same hands (B1-serum vs B1-heparin)
  and obtained multiple peptides for many proteins
  which had just one hit with LCQ.

28
SPECIFIC OBSERVATIONS DEPLETION

• Many investigators depleted albumin and/or
  immunoglobulins
• Several were provided Agilent immunoaffinity
  column to remove top-6 proteins
• Much higher numbers of identifications after
  depletion if sufficient fractionation
• Inadvertent removal of other proteins sponge
  effect of albumin
• Assay both flow-through bound fractions

29
SPECIMEN VARIABLES

• What evidence have we developed for choice of
  specimens for analysis?
• Plasma preferred over serummore consistent, less
  degradation
• EDTA-plasma preferred over heparin interferences
  and citrate dilution
• Clot activator? necessary only for serum
• Minimize freeze/thaw cycles (archives)
• Minimal evidence of platelet activation 4C
• Protease inhibitors desirable, but alter proteins

30
INFLUENCE OF ABUNDANCE

• Using quantitative immunoassays and microarrays
  (generally unknown epitopes), we have found very
  high rates of detection of the more abundant
  proteins, less in the mid-range, and occasional
  detection of very low abundance proteins, as
  expected.
• High correlation (r0.9) between peptides and
  measured concentrations

31
Least Abundant Proteins Identified with two
distinct peptides(pg/ml range 200 pg/ml to 20
ng/ml)

• Alpha fetoprotein
  2.9E-02
• TNF-R-8
  3.3E02
• TNF-ligand-6
  1.5E03
• PDGF-R alpha
  4.6E03
• Leukemia inhibitory factor receptor 5.0E03
• MMP-2/gelatinase
  8.8E03
• EGFR
  1.1E04
• TIMP-1
  1.4E04
• IGFBP-2
  1.5E04
• Activated leukocyte adhesion mol 1.6E04
• Selectin L five labs10 peptides
  1.7E04

32
BIOLOGICAL INSIGHTS

• The proteins identified can be annotated by many
  methods. We have searched multiple databases,
  including Gene Ontology, Novartis Atlas, Online
  Mendelian Inheritance in Man (OMIM), incomplete
  or unidentified sequences in the human genome,
  microbial genomes, InterPro protein domains,
  transmembrane domains, secretion signals.
• See Proteomics 2005 53226-3519 Wiley, 2006

33
GENE ONTOLOGY SPECIFIC TERMS

• Over-represented in PPP 3020 (vs whole genome)
  extracellular, immune response, blood
  coagulation, lipid transport, complement
  activation, regulation of blood pressure, as
  expected also cytoskeletal proteins, receptors
  and transporters.
• Proteins from most cellular locations and
  molecular processes are recognized.
• Under-represented perception of smell (1 vs
  25 exp) cation transporters, ribosomal proteins,
  G-protein coupled receptors, and nucleic acid
  binding proteins.

34
InterPro Protein Domain Analysis

• Compared with the whole human genome, the 3020
  PPP proteins are
• Over-represented for EGF, intermediate filament
  protein, sushi, thrombospondin, complement C1q,
  and cysteine protease inhibitor.
• Under-represented Zinc finger (C2H2, B-box,
  RING), tyrosine protein phosphatase, tyrosine and
  serine/threonine protein kinases,
  helix-turn-helix motif, and IQ calmodulin binding
  region domains.

35
TRANSMEMBRANE AND SECRETED PROTEIN FEATURES

• 1297 of 3020
• SwissProt Annotated
  ProFun Both
• Transmembrane 230 151 104
• Secretion signal 373 420
  358
• 1723 of 3020
  ProFun Predicted
• TM domain(s)
  137
• Secretion signal
  255

36
Cardiovascular-Related Proteins Biomarker
Candidates in the PPP Database

• Proteins characterized in eight groups
• Inflammation
• Vascular
• Signaling
• Growth and differentiation
• Cytoskeletal
• Transcription factors
• Channels
• Receptors

37
Comparison of Five Search Algorithms

• Using PPP data, Kapp et al (Proteomics 2005)
  found Sequest and Spectrum Mill more sensitive
  and MASCOT, Sonar, and X!Tandem more specific for
  peptide identifications at specified
  false-positive rates.
• Some investigators have reported using
  combinations of two or more search engines.
  Decision rules are necessary.

38
Can We Overcome the Idiosyncrasies of Individual
Instruments and Laboratories?

• Several informatics investigators approached the
  human PPP with an offer to re-analyze the
  complete MS/MS datasets using their own software
  and criteria from the raw spectra (or peaklists).
• These analyses eliminated the heterogeneity of
  search algorithms, search parameters, and
  idiosyncrasies of individual labs.
• The results are hard to compare, given different
  extent of analysis. However, each can be
  compared with the Core Dataset.

39
Independent Analyses from Raw Spectra (IDs with
2 peptides)

• Core Dataset (18 datasets, 3020)
• PepMiner (Beer, 8 large datasets, 2895) 1051 in
  3020 dataset, 700 in the 9504
• X!Tandem (Beavis/States, 18 datasets, 2678) 577
  in the 3020 218 in the 889
• PeptideProphet/ProteinProphet (Deutsch, 7
  datasets, 960)479 in 3020
• Mascot/Digger (Kapp, Australia, 14 datasets, 513
  with 1.4 error rate ongoing analysis

40
What is Required and Feasible to Enhance the
Statistical Robustness of Findings?

• Many complex proteomics analyses are done once,
  without replicates required to estimate
  coefficient of variation or other standard
  parameters for clinical chemistry use.
• Five to ten independent repetitions of the
  experiments are a must Hamacher et al,
  Proteomics in Drug Discovery, 2006.
• How should we determine how similar or different
  are samples A and B, or the results of methods X
  and Y? What decision rules apply?
• We have a long way to go from discovery research
  to clinical applications.

41
Comparison of 5 Published Reports on Plasma
Proteins with HUPO PPP Datasets

• Report IDs IPI in 3020 in 9504
• Anderson 1175 990 316 471
• Shen 1682 1842 213 526
• Chan 1444 1019 257 402
• Zhou 210 148 51 88
• Rose 405 287 142 159

42
Comparison of New Biofluid Proteome Findings with
HUPO PPP-3020 Proteins

• Proteome Proteins IPI 2.21 PPP-3020
• Urine 1543 910
  293
• tears 491 313
  117
• semen 923 560
  180
• Refs from Matthias Mann Lab, Genome Biology,
  2007, different IPI versions.
• Comparison, Omenn, Proteomics-Clinical
  Applications (2007).

43
NEXT PHASE OF PPP (PPP-2)

• Standard operating procedures (SOPs), including
  EDTA-plasma as standard specimen replication and
  confirmation of results
• Quantitation and subproteomes, using new methods
  and advanced instruments
• Databases and robust bioinformatics
• Clinical chem/disease-related studies

44
PPP-2 Research Technology Thrusts

• Learned a lot from Pilot Phaseplasma is a very
  complex specimen no single platform sufficient
  analyses currently far from comprehensive, let
  alone reproducible now have improved data
  quality and informatics resources.
• PPP-2 use multiple methods focus on biomarker
  discovery build upon already-funded laboratories
  and repositories.

45
Specific Technology Recommendations

• N-Glycosite (proteotypic) peptide resource is
  a special subproteome likely to have high
  biomarker relevance.
• Capture glycoproteins, digest with
  trypsin and PNGase F to yield N-linked
  glycopeptides. Choose one unique to each protein
  a finite number not all proteins. Use
  complementary lectin approach to characterize
  glycans.
• Prepare isotope-labeled
  N-glyco-peptides for multiple uses as standards
  and to spike specimens.

46
N-Glycosites
Glycoproteins are enriched on cell surface, in
secreted proteome and in plasma Glycoproteins
tend to be stable Only few glycosites per
protein reduction in sample complexity (excludes
albumin) Inherent validation of N-glycosite by
fragment ion spectrum N-glycosite subproteome is
probably the one easiest to completely map
47
Glycopeptide Isolation
Zhang H., Li X.-J., Martin D.B. Aebersold R.
(2003) Nat Biotech 21 660-666
48
Flow chart of process
Tissue Samples
Plasma Samples
Normal Disease
Capture / Digestion
'Glycopeptide' Fract.
'Glycopeptide' Fract.
LC-MS
LC/MS Maps
Target peptides
Data Analysis
MRM LC/MS/MS
Data Analysis
Targeted LC/MS/MS
49
Reducing Complexity Glycoprotein-Enriched
Subproteomes

• Methods Lab 2
  Lab 11
• Enrichment hydrazide chem lectin chromy
• Peptide Fxn SCX RP RP
• Mass Spec qtof
  deca-xp
• Search engine Seq/ProteinProphet Sequest
• Protein IDs 222
  83
• in B1-serum 51 in common
• Of total 254, 164 found among data from 11 other
  labs without glycoprotein enrichment.

50
Technology Recommendations (contd)

• Orbitrap and other advanced instruments with high
  mass accuracy and increased throughput
• Multiple Reaction Monitoring (Q-Trap, triple
  quad---LOD lt50 amol, 5 logs range, probably ng/ml
  range for GP.
• Extensive fractionation and newer labeling
  methods.
• Recruit several major labs be open to
  volunteers.
• Determine interest in reference specimen.
• Make peptide standards available through PPP-2
  post lists and make labeled compounds.

51
Multiple Reaction Monitoring (MRM)

• High selectivity two levels of mass selection
  (increased S/N)
• High sensitivity because of high duty cycle (Q1
  and Q3 are static)
• Only known peptides (candidates) are detected

52
Technology Recommendations (contd)

• Compare pooled samples from disease and control
  high throughput not essential for discovery phase
• Continue to build the catalog
• Do longitudinal repeat measures on individuals to
  establish CoVmust reliably tell whether two
  samples are the same or different, including PTMs
• Pay attention to precursor ions
• Known interested labs Aebersold, Paik, Smith,
  Speicher, Hancock, Mann probably Chinese,
  Michigan, FHCRC, Japanese/glycomics.

53
Issues for PPP Bioinformatics

• What are imperatives for project design?
• How can many more spectra be interpreted?
• How can more confident protein IDs be generated?
• How do we add value and benefit from EBI/PRIDE
  and ISB/PeptideAtlas repositories?
• What is required to make the datasets more useful
  for other investigators?
• Can quantitation, including of PTMs, be achieved
  with statistical robustness?

54
A Robust Bioinformatics Architecture
PRIDE
Dissemination
Peptide Atlas
Genome annotation
Level I repository
Individual labs
55
Repositories and Resources for Proteomics
Informatics

• PRIDE at EBI, repository for protein
  identifications (Martens)
• PeptideAtlas, repository for raw data processed
  through TransProteomics Pipeline at ISB
  (Deutsch), plus SpectraST barcodes from NIST
• Tranche Distributed File System/DFS (Andrews, UM)
  at ProteomeCommons.org, National Resource for
  Proteomics and Pathways
• CPAS, developed as part of Mouse Models of Human
  Cancers Consortium, at Fred Hutchinson (McIntosh)
• GPMdb, developed by Beavis (Canada)

56

• Tranche Distributed (P2P) File System
• Open, simple, cross-platform protocols
• e-Commerce-grade encryption makes it appropriate
  for scientific research (peer-review and
  traceability)
• Can easily grow to accommodate very large amounts
  of data and users
• Commodity hardware _at_ 0.37 per GB storage
• 16 TB over 12 servers (30 additional TB ordered)
  and funding for additional 20TB
• Documentation, tools, code, credits
  http//www.proteomecommons.org/dev/dfs
• Data sets GPM, PNNL, Aurum, QqTOF vs QSTAR, sPRG
  ABRF 2006, HUPO PPP
• Links with PeptideAtlas, OPD, HPRD, TheGPM

57
Can We Identify More High Confidence Peptides
from the MS/MS spectra?

• The spectra, not protein lists, are the raw data.
  lt20 of spectra are confidently assigned to
  peptide sequences the rest are typically
  discarded.
• More high quality spectra can be mined
  (Nesvizhskii et al, MCP 2006).
• Higher mass accuracy greatly enhances results
  (with some complications---Eric Deutsch).
• Error estimates and thresholds should be routine
  for peptide IDs and protein matches.
  TransProteomicPipeline (TPP) from ISB has been
  designed for this purpose.

58
Mining Un-assigned High Quality
Spectra(Nesvizhskii)

• Typical search SEQUEST, IPI database
• semi-constrained (tryptic on one end)
• Met 16
• /- 3 Da, average mass
• Average numbers (LCQ/LTQ data) 10-15 of all
  spectra assigned peptide with high confidence
• 20-25 of all high quality spectra are not
  assigned

59
Why Are Spectra Not Assigned?

• Possible causes of failure to assign peptide
• Imperfect scoring scheme
• Constrained search (PTM, not tryptic etc.)
• Incorrect mass/ charge state
• Low spectrum quality / contaminant ion
• Correct sequence may not be in the database
  searched (e.g., SNP)
• Novel sequence (splice variants, fusion
  peptides?)
• Use MS/MS data for genome
  annotation

60
Finding and Mining High Quality Unassigned
Spectra (Nesvizhskii)
61
Further Analyses at the Peptide Level

• The PPP, GPM, and PeptideAtlas databases are rich
  with peptide-level findings, which can be
  analyzed for many questions---e.g., which
  peptides are most likely to be detected from
  among the predicted tryptic peptides of various
  proteins, and why? Can peptides be used directly
  to identify sequences of splice isoforms and
  SNPs? Can PTMs be identified more readily?
  Answers Yes to all three questions.
• Proteotypic peptides will be a major feature of
  Next Phase PPP.

62
What Kinds of Biological Insights Emerge from
Annotation?

• The aim of proteomics analyses is not just to
  create lists of peptides and proteins, but to
  advance our understanding of complex biological
  processes in health and disease. Going forward,
  quantitation of proteins and their PTMs will be
  increasingly important---and feasible.

63
High Throughput Proteomics and Systems Biology

condition 1
Understanding and modeling cell behavior Systems
Biology
condition 2
condition 3
Integration of genomic, transcriptomic,
proteomic, metabolomic data
64
SUMMARY

• Enthusiasm for continuing and expanding Plasma
  Proteome Project, confirmed at Seoul, Korea,
  World Congress of Proteomics Oct 2007
• Commitment to combine PPP with concept of Disease
  Biomarker Initiative
• Interest in linking with and absorbing datasets
  from other Biofluid Proteomes (saliva, urine,
  CSF, organ-related proximal fluids)

65
Biology as an Information Science NIH Roadmap
National Centers for Biomedical Computing
Informatics for IntegratingBiology and the
Bedside (i2b2) Isaac Kohane, PI
Physics-Based Simulation of Biological Structures
(SIMBIOS) Russ Altman, PI
National Center for Integrative Biomedical
Informatics (NCIBI) Brian Athey, PI
National Alliance for Medical Imaging Computing
(NA-MIC) Ron Kikinis, PI
The National Center For Biomedical Ontology
(NCBO) Mark Musen, PI
Multiscale Analysis of Genomic and Cellular
Networks (MAGNet) Andrea Califano, PI
Center for Computational Biology (CCB) Arthur
Toga, PI
66
A Bioinformatics Approach to Discover Candidate
Oncogenes

• Few causal cancer genes have been discovered
  using gene expression microarrays
• Oncogenic events are often heterogeneous
• ERBB2/HER2 amplification in 20 of breast CA
• Activating Ras mutations in 25 of melanomas
• E2A-PBX1 translocation in 5-10 of leukemias
• Chromosomal aberrations that result in marked
  over-expression of an oncogene should be
  detectable in transcriptome data
• Protein products then may be identified in tumor,
  biofluids, and plasma

67
(No Transcript)
68
COPA of microarray data revealed ETV1 and ERG as
outlier genes across multiple prostate cancer
gene expression data sets
Tomlins et al., Science 2005, 310 644
-648
69
COPA Unveils Androgen-Responsive TF Fusion
Genes
70
The Molecular Concept Map Project Chinnaiyan,
Rhodes
71
(No Transcript)
72
Our Genetic Future

• Mapping the human genetic terrain may rank with
  the great expeditions of Lewis and Clark, Sir
  Edmund Hillary, and the Apollo Program.
  --Francis Collins, Director
• National Human Genome
  Research Institute, 1999
• Next
• Understand gene and protein expression
• Elucidate genetic, environmental, and
  behavioral interactions in health and disease
• Engage scientists globally

73
Acknowledgements

• HUPO PPP Ruedi Aebersold and Young-Ki Paik,
  co-chairs Eric Deutsch, Lennart Martens, Alexey
  Nesvizhskii, David States, bioinformatics lab
  leaders and sponsors (see Proteomics 2005)
• UM Proteomics Alliance for Cancer Research Phil
  Andrews, David States, Alexey Nesvizhskii, George
  Michailidis, Mike Pisano, Arul Chinnaiyan, Dan
  Rhodes, Scott Tomlins, Arun Sreekumar, Adai
  Vellaichamy, Brian Haab
• UM National Center for Integrative Biomedical
  Informatics Brian Athey, David States, HV
  Jagadish, Jignesh Patel, Peter Woolf, Biaoyang Lin