Protein Mutations and Pathways in Cancer Toward Modular

1
Protein Mutations and Pathways in CancerToward
Modular Combinatorial Therapy
Chris Sander Computational Systems
BiologyMemorial Sloan-Kettering Cancer Center,
New York
International Conference on BioinformaticsAsia-Pa
cific Bioinformatics Network
Less war !!
more science !!
2

• Cancer
• Simplicity of phenotype
• Diversity of implementation
• Modular therapy!
• Combinatorial therapy!

3

• Cancer Genomics
• Functional consequences of somatic mutations
• Molecular alterations in pathway context
• Toward Combinatorial Therapy
• Combinatorial Perturbations Network Models
• Information Infrastructure
• Pathway Commons Author Fact Deposition

4
Cancer GenomicsNikolaus SchultzBarry Taylor,
Ethan CeramiNick Socci, John MajorSam
SingerMarc Ladanyi, Cameron BrennanMatt
Meyerson, Jordi Barretina TCGA community
Niki Schultz

Function of Protein MutationsBoris RevaJenya
Antipin Alyosha Stupalov
PathwayCommons.org Emek DemirGary Bader,
Toronto Ethan CeramiBen GrossRobert
HoffmanKen FukudabioPAX community
5
The Cancer Genome Atlas (TCGA)
Sample processingClinical annotation
DNA copy number
DNA methylation
Gene expression - mRNA (exon-level) - miRNA
DNA sequencingcurrently 1300 genes soon 6000 or
all genes
Genomic rearrangements Proteomics
Data storage and distribution
Integrative analysis
Next lung squamous, kidney, breast, colon
6
Glioblastoma Multiforme Aggressive Disease
7
Complexity and diversity of genetic alterations
in cancer
Many samples High resolution
Individual samples Low resolution
TCGA - GBM
8
DNA copy number alterations in GBM and ovarian
cancer
DNA copy number
DNA methylation
mRNA expression
miRNA expression
mutations
Clinical data
OVA
GBM
More than half of the genes copy-number altered
in ovarian cancer correlated with expression
9

• Cancer Genomics
• Functional Consequences of Protein Mutations

10
Somatic mutations in cancerWhat are the
functional consequences ?
Input
allele 1 GCC ATC CCG
ALA ILE/MET PRO allele 2 GCC AAC
CCG
mutation in coding region
Public Databases
protein family
3D structure
3D complex
pathway
Q9SD07 RLHIGGL Q44861 RLLIGRV Q61EI4
RLLIGRV Q7PY36 RLFIGKI Q4I4E0 RLWMDQI Q58SJ9
RLLIGRV O01811 RLFIGKI P55159 RLLIGRV
Superfamily PDB PFAM SCOP NCBI ENSEMBL Reactome
protein stability
correlation between interacting residues
specificity conservation
protein-protein interactions
pppi
psc
pcr
pps
Output
Probability(Disruptive/Non-disruptive) f (
Psc, Pcr, Pps, Pppi )
11
Assessing the functional consequences of mutations
EGFR_human
12
CEO algorithmCombinatorial Entropy Optimization

• Boris Reva, MSKCC

13
Defining subfamilies and specificity residues
Input
Output
1
2
Clustering
Sub-Families
3
4
Specificity Residues
Conserved Residues
14
Q How one can achieve the most
distinctiveinformative separation of sequences
into clusters?
Minimize contrast function difference between
entropies of ordered and disordered clusters of
sequences of the same size
S
ordered
S
disordered
S-S-9
S-S0
S-S-3.5
S-S-7.5
Goal S-S-gtmin
15
combinatorial entropy measure of specificity
patterns
is the number of sequences in cluster
(subfamily) k is the number of
residues of type a in the column i of the
cluster k.
For each column i of the alignment one computes
the combinatorial entropy and the reference
entropy
The entropy difference , summed
up over all columns i, is a measure of the
deviation of a given sequence clustering from
random. This difference is minimal when each
cluster has its distinct type of residues.
Optimization problem form clusters (subfamilies)
of sequences, so as to minimize the combinatorial
entropy difference .
16
Specificity residues - high contrastGlobally
conserved residues - low contrast
Family of 390 protein kinases
17
Defining subfamilies and specificity residues
Input
Output
1
2
Clustering
Sub-Families
3
4
Specificity Residues
Conserved Residues
18
Assessing the functional consequences of mutations
EGFR_human
19
OMA - Online Mutation Analysis
www.cbio.mskcc.org/cancergenomics
www.proteinfunction.org
20
Functional implications of cancer mutationsat
the protein level ERBB2 mutationsL49H no
alignment data available C311R strong functional
impact, conserved residue N319D likely
functional impact, conserved and specificity
residue E321G likely functional impact,
specificity residue D326G likely functional
impact, specificity residue - binding
site? C334S strong functional impact, conserved
residue in S-S bridge V750E strong functional
impact, strongly conserved residue V777A unlikely
functional NF1 mutations V1308E strong
functional impact, buried residue R1412S strong
functional impact D1849N no alignment data
available A2336T likely functional impact,
specificity residue
21
Examples of mutations predicted as functional by
OMA
D326G in ERBB2- Tyrosine kinase-type cell surface
receptor HER2
D-gtG
22
Examples of mutations predicted as functional by
OMA
C334S in ERBB2 - Tyrosine kinase-type cell
surface receptor HER2
strong functional impact conserved
residue mutation eliminates SS bridge C334-C338
C334
C338
23
Networks of genes correlated by co-occurrence of
mutations
Connected genes are observed simultaneously
mutated in at least 3 tumor samples probability
of co-occurrence of these mutations by chance
(P-value) is less than 0.005. Reva et al., Sander
group, cBio_at_MSKCC
24

• Cancer Genomics
• Molecular alterations in pathway context

25
TCGA glioblastomaDNA copy number changes and
mutations
amplified
mutated
deleted
26
Glioblastoma copy number alterationsWhich events
are functional, which are passengers ?
RAE recurrence amplitude extent
RAE Barry Taylor, Nick Socci, Chris Sander PLoS
ONE 2008
27
Analyzing genetic alterations in pathway context
www.cbio.mskcc.org/cancergenomics
28
Combining molecular profiles and prior biological
knowledge
www.cbio.mskcc.org/cancergenomics
29
GBM pathway
Based onGenes Dev. 2007 Nov 121(21)2683-710.
30
copy number datasample 2
31
copy number datasample 3
32
copy number datasample 4
33
copy number datasample 5
34
copy number datasample 6
35
copy number datasample 7
36
copy number datasample 8
37
mRNA expr. datasample 8
38
methylation datasample 8
39
mutation datasample 8
40
mutation datasample 3
41
Mapping molecular alterations in 200 glioblastoma
samples onto biological pathwaysGoal determine
oncogenic programs
www.cbio.mskcc.org/cancergenomics
42
Cancer program by sub-networks
The CancerGenome AtlasPilot Project(2006-2008)
200 cases ofglioblastoma m. brain tumors
www.cbio.mskcc.org/cancergenomics
43
Automate module analysis (make it objective)
44
Key capture biological knowledge in computable
form
http//www.pathwaycommons.org
bioPAX
Facilitate creation and communication of pathway
data Aggregate pathway data in the public
domain Provide easy access for pathway analysis
Community Process !
45
Algorithm(s) to detectaltered modules in cancer
46
glioblastoma altered modules
47
PI3K module change in subtypes ?
48
whole proteome/genome sequencing will lead to
more complete module map
49
Cancer network module analysis - nextmissing
measure protein modification levelsmissing
histone etc. modificationsadd 6K 20K protein
seqsspecificity distribution of
interactionstime progression (dynamics)tumor
subtypescell types within tumor sample
50
Theoretical framework wanted

• Specificity rank distribution for protein
  domains !
• Alterations of specificity distribution in
  cancer
• reduction
• expansion
• switch

51
Cytoscape, bioPAX Pathway Commons Emek
DemirEthan CeramiBen GrossRobert HoffmanKen
FukudabioPAX communityGary BaderPerturbational
cell biologySven Nelander Wei Qing Wang Peter
GennemarkNeal Rosen, Christine Pratilas
Cancer GenomicsNikolaus SchultzBarry
TaylorBoris Reva, J Antipin, A StukalovJohn
MajorNick Socci, Sam SingerMarc LadanyiMatt
Meyerson, Jordi Barretina
Small RNAsDoron BetelRob Sheridan Christina
Leslie, Debora MarksTom Tuschl, Eric Kandel
Protein Families Combinatorial EntropyBoris
Reva, Jenya Antipin

TGFbeta 6000 gene RNAi screenNikolaus
Schultz Dina Marenstein Joan Massague, Hakim
Djaballah
Support Bioinformatics Core in the
Computational Biology Center at MSKCC
tools gt
52

• Network pharmacology
• Toward Combinatorial Therapy
• Simple Models from Complex Data
• CoPIA Nelander et al. - 2008

53
Perturbation Cell Biology CoPIASven
NelanderPeter Gennemark Wei Qing Wang
Bjoern Nilsson, Christine Pratilas, QingBai
SheNeal Rosen

Sven Nelander
http//cbio.mskcc.org/copia/ Nelander, Sander et
al., Molecular Systems Biology, 2008
54
Reality
Abstraction / Model
Application
Therapy
55
Experiment Dual drug perturbation of MCF7 cancer
cell line_at_ MSKCC
Wei Qing Wang, Sven Nelander Rosen Lab
2007-2008
56
Mathematical Model System Simulation by Bounded
ODEslike Hopfield Network
linear
non-linear
Mean Field Model for Combinatorial Perturbation
57
A simple but effective non-linear deviceto
capture cooperative effects(epistatis, synergy,
antagonism)
transfer function f()
58
Optimize the network model

• Minimize the discrepancy between prediction and
  experiment
• while keeping the model simple !

Sum of squares pred-expt error
Structural complexity
59
Optimization algorithm examplesexplore
alternative network structuresmovie by Evan
Molinelli
60
Result Network Model(s)
61
Dual drug perturbation in MCF7 cancer cell line_at_
MSKCC
62
Best network model deduced from dual drug
perturbation in MCF7 cancer cell lines
63
Dual drug perturbation in MCF7 cancer cell line_at_
MSKCC
Does the model work ? Leave out one drug combo at
a time, compute best model, predict compare
with experiment
64
Set of Best Network Model(s)
65
Interactions deduced from dual drug perturbation
from a set of best network models in MCF7 cancer
cell lines
66
Network structure deduced from dual drug
perturbation in MCF7 cancer cell lines
67
Network structure deduced from dual drug
perturbation in MCF7 cancer cell lines
68
EGF
IGF-1
EGFR
PKC-?
IGF1R
RAF
IRS-1
PI3K
MEK
direct activation
TSC1/2 complex
Indirect or unknown mechanism
AKT
ERK
direct inhibition
RheB
Cyclin D1
GSK3?/?
mTOR
4E-BP1
p70S6K
S6
eIF4E
29.3.2008 WeiQing Wang
69
Power of CoPIA network models CoPIA
Combinatorial Perturbation Analysis
Capture multiple perturbation epistasis
(synergy/antagonism) feedback loops time-dependent
processes modification of prior knowledge
70
CoPIA Network Models - Applications
refine pathway models
identify drug targets
design combination therapy
predict outcomes
71
How do we influence cancer cells ?
Inhibit anti-apoptotic signals Inhibit growth
signals Re-differentiate
Re-configure cellular networks !
New developments Multiple agents Rational
time sequence Genotype specific therapy
72
Extend complete existing network models !
73

• Cancer Genomics
• Functional consequences of somatic mutations
• Molecular alterations in pathway context
• Toward Combinatorial Therapy
• Combinatorial Perturbations Network Models
• Information Infrastructure
• Pathway Commons Author Fact Deposition

74
(No Transcript)
75
Integrate Pathway Information
http//www.pathwaycommons.org
bioPAX
Facilitate creation and communication of pathway
data Aggregate pathway data in the public
domain Provide easy access for pathway analysis
Community Process !
76
(No Transcript)
77
http//iHOP-net.orgGenes compounds
interactions from millions of abstracts -
instantly
Robert Hoffmann, Benjamin Gross, Chris Sander
iHop-net.org version 2 released 6 Dec 2006
78
Factoidsdigital abstracts to databases
How to get rich biological knowledge into a
computable form

• As authors submit a paper they deposit
  structured facts to a public database

79
Postdocs wantedSander Group Computational
Systems Biology _at_ MSKCC in NYCUpper East Side
Tri-I Campus Sloan Kettering, Cornell Weill,
Rockefeller
We pause for station identification

• Cancer genomics (dry)
• Network pharmacology (wet)

80
Summary
Summary

• Toward Combinatorial Therapy
• Use multiple perturbation experiments to build
  predictive network models
• Cancer Genomics
• The active sub-pathway model of cancer biology
• Pathway Commons
• One-stop-shop access to pathway informationusing
  the bioPAX common language

81
Cytoscape, bioPAX Pathway Commons Emek
DemirEthan CeramiBen GrossRobert HoffmanKen
FukudabioPAX communityGary BaderPerturbational
cell biologySven Nelander Wei Qing Wang Peter
GennemarkNeal Rosen, Christine Pratilas
Cancer GenomicsNikolaus SchultzBarry
TaylorBoris Reva, J Antipin, A StukalovJohn
MajorNick Socci, Sam SingerMarc LadanyiMatt
Meyerson, Jordi Barretina
Small RNAsDoron BetelRob Sheridan Christina
Leslie, Debora MarksTom Tuschl, Eric Kandel
Protein Families Combinatorial EntropyBoris
Reva, Jenya Antipin

TGFbeta 6000 gene RNAi screenNikolaus
Schultz Dina Marenstein Joan Massague, Hakim
Djaballah
Support Bioinformatics Core in the
Computational Biology Center at MSKCC
tools gt
82
Optimization algorithm 1outer loop - explore
alternative network structures
error
model
occasoinally climb uphill in Monte Carlo fashion
83
Optimization algorithm 2inner loop optimize
parameters along a trajectory in solution space
model
error
adjust weights
explicit
result optimal model parameters Wij