Keynote lecture for NetBio SIG 2012

1. Predicting the impact of
mutations using pathway-
guided integrative genomics
Network Biology SIG, ISMB 2012
Josh Stuart, UC Santa Cruz
July 12, 2012

2. Overview of pathway-guided approach
Integrate many data sources to gain accurate
view of how genes are functioning in pathways
Predict the functional consequences of mutations
by quantifying the effect on the surrounding
pathway
Use pathway signatures to implicate mutations in
novel genes to (re-)focus targeting
Identify critical “Achilles Heels” in the pathways
that distinguish a particular sub-type

3. Flood of Data Analysis Challenges
Genomics, Functional Genomics, Metabolomics, Epigenomics =
Exome
Sequences
Multiple, Possibly
Structural Conflicting Signals
Variation Expression
Copy Numberis What it
This
Does to You
Alterations DNA Methylation

4. Analysis of disease samples like automotive repair
(or detective work or other sleuthing)
Patient Sample 1 Patient Sample 2
Sleuths use as much
knowledge as possible.
Patient Sample 3 Patient Sample N
…

5. Much Cell Machinery Known:
Gene circuitry now available.
Curated and/or Collected
Reactome
KEGG
Biocarta
NCI-PID
Pathway Commons
…

6. 

Expression of 3 transcription factors:
high TF
high TF low TF
Inference: Inference: Inference:
TF is ON TF is OFF TF is ON
(expression (high expression (low-expression
reflects but inactive) but active )
activity)

7. 
BUT, targets are amplified
Expression -> TF ON Copy Number -> TF OFF
TF
Lowers our belief
in active TF because
explained away by
cis evidence.

8. Nir Friedman, Science (2004) - Review






9. Integration Approach: Detailed models of
gene expression and interaction

10. Integration Approach: Detailed models
of expression and interaction
Two Parts:
1. Gene Level Model
(central dogma)
2. Interaction Model
(regulation)

11. 1. Central Dogma-Like
Gene Model of Activity
2. Interactions that
connect to specific points
in gene regulation map
Charlie Vaske
Vaske et al. 2010. Bioinformatics Steve Benz

12. Multimodal Data Pathway Model
Cohort Inferred Activities
of Cancer
CNV
mRNA
meth
…




13. Patient Samples (247)
Pathway Concepts (867)
TCGA Network. 2011. Nature
(lead by Paul Spellman)

14. Patient Samples (247)
FOXM1 Transcription Network
Pathway Concepts (867)
TCGA Network. 2011. Nature
(lead by Paul Spellman)

15. TCGA Network. 2011. Nature (lead by Paul Spellman)

16. Mutated genes are the focus of many targeted
approaches.
Some patients with “right” mutation don’t respond.
Why?
Many cancers have one of several “novel” mutations.
Can these be targeted with current approaches?
Pathway-motivated approaches:
 Identify gain-of-function from loss-of-function.
Sam Ng
 Compare novel signatures ISMB
Oral Poster

17. High
Inferred
Activity Inference using Inference using
Inference using all downstream upstream neighbors
neighbors neighbors
mutated
FG gene FG SHIFT FG
Low
Inferred
Activity
Sam Ng, ECCB 2012

18. FG
Sam Ng, ECCB 2012

19. 1.
Identify
FG Local FG
Neighbor-
hood
Sam Ng, ECCB 2012

20. 2a.
Regulators
Run
1. FG
Identify
FG Local FG
Neighbor- 2b.
hood Targets FG
Run
Sam Ng, ECCB 2012

21. 2a.
Regulators
Run 3.
1. FG
Calculate
Identify
FG Local FG FG - FG P-Shift
Score
Neighbor- 2b. Difference
hood Targets FG
FG
Run
(LOF)
Sam Ng, ECCB 2012

22. Shift Score
PARADIGM downstream
PARADIGM upstream
Expression
Mutation
RB1
Sam Ng, ECCB 2012

23. Focus Gene Key
P-Shift
T-Run
R-Run
Expression
Mutation
Neighbor Gene Key
Activity
Expression
RB1 Mutation
Sam Ng, ECCB 2012

24. RB1 Discrepancy Scores distinguish
mutated vs non-mutated samples
Signal Score (t-statistic) = -5.78
Sam Ng

25. 


Observed SS
Background SS
Sam Ng

26. TP53 Network
Sam Ng

27. P-Shift Score
PARADIGM downstream
PARADIGM upstream
Expression
Mutation
NFE2L2
Sam Ng

28. Focus Gene Key
P-Shift
T-Run
R-Run
Expression
Mutation
Neighbor Gene Key
Activity
Expression
RB1 Mutation
NFE2L2
Sam Ng

29. RB1 TP53 NFE2L2
Signal Score (t-statistic) = -5.78 Signal Score (t-statistic) = -10.94 Signal Score (t-statistic) = 4.985
Observed SS
Background SS
Sam Ng

30. 

“ ”



 “ ”
’
Sam Ng

31. Sam Ng

32. Pathway Discrepancy HIF3A (n=7)
TBC1D4 (n=9) (AKT signaling)
NFE2L2 (29)
MAP2K6 (n=5)
LUSC
MET (n=7) (gefitinib resistance)
GLI2 (n=10) (SHH
signaling) CDKN2A (n=30)
AR (n=8)
EIF4G1 (n=20)




 Sam Ng

33. Christina Yau,
Buck Inst

34. Christina Yau, Buck Inst.

35. Identify sub-pathways that
distinguish patients sub-types (e.g. Insight from contrast
mutant vs. non-mutant, response
to drug, etc)
Predict mutation impact on
pathway “neighborhood”
Identify master control points for
drug targeting.
Predict outcomes with quantitative
simulations.
Sam Ng Ted Goldstein

36. “ ”
Pathway Activities
Pathway Activities
Ted Goldstein Sam Ng

37. “ ”
SuperPathway Activities
SuperPathway Activities
Pathway
Signature
Ted Goldstein Sam Ng

38.  Traditional methods treat
each gene as a separate
feature
 Use features reflecting
overall pathway activity
 Smaller number of
features are now fed to
predictors
Predictor
Artem Sokolov

39.  Traditional methods treat
each gene as a separate
feature
 Use features reflecting
overall pathway activity
 Smaller number of
features are now fed to
predictors
Artem Sokolov
Predictor ISMB poster

40. Basal vs.
Luminal
Recursive
feature
elimination:
we train an
SVM, drop the
least
important half
of features
and recurse
The number of
times each
feature
survived the
elimination
across 100
random splits
of data
Artem Sokolov

41. Methotrexate
Sensitivity
Non sub-type
specific drug
Pathway involving
the target of the
drug.
Artem Sokolov

42. One large highly-connected
component (size and connectivity
significant according to permutation
980 pathway concepts
analysis)
1048 interactions
Characterized by
several “hubs’
IL23/JAK2/TYK2
P53 ER
tetramer
HIF1A/ARNT
FOXA1
Myc/Max
Higher activity in ER-
Lower activity in ER-
Sam Ng, Ted Goldstein

43. Identify master controllers using
SPIA (signaling pathway impact analysis)
Google PageRank for Networks
Determines affect of a given pathway on each node
Calculates perturbation factor for each node in the network
Takes into account regulatory logic of interactions.
n
IF ( g j )
Impact IF ( gi ) = s ( gi ) + å bij ×
factor:
j=1 N up ( g j )
Google’s PageRank-Like
Yulia Newton (NetBio SIG Poster)

44. Slight Trick: Run SPIA in reverse
Reverse edges in Super Pathway
High scoring genes now those at the “top” of the
pathway
PageRank finds Reverse to find
highly referenced Highly referencing
Yulia Newton

45. Master Controller Analysis on Breast Cell
Lines
Basal
Luminal
Yulia Newton

46. • DNA damage network is
upregulated in basal
breast cancers
• Basal breast cancers are
sensitive to PLK inhibitors
GSK-PLKi
Basal
Claudin-low
Luminal
Ng, Goldstein Up
Heiser et al. 2011 PNAS Down

47. • HDAC Network is down-
regulated in basal breast
cancer cell lines
• Basal/CL breast cancers are
resistant to HDAC inhibitors
HDAC inhibitor VORINOSTAT
Heiser et al. 2011 PNAS Ng, Goldstein

48. Connect genomic alterations to downstream
expression/activity
?
• What circuitry connects mutations to
transcriptional changes?
– Mutations  general (epi-) genomic perturbation
– Expression  activity
• Mutation/perturbation and expression/activity
treated as heat diffusing on a network
– HotNet, Vandin F, Upfal E, B.J. Raphael, 2008.
Evan Paull
– HotNet used in ovarian to implicate Notch pathway
• Find subnetworks that link genetic to mRNA and ISMB
protein-level changes. Oral Poster

49. HotLink
Genomic Perturbations Gene Activity
(Mutations, Methylation, Focal Copy Number) (Expression, RPPA, PARADIGM)
?
Evan Paull

50. HotLink
Genomic Perturbations Gene Activity
(Mutations, Methylation, Focal Copy Number) (Expression, RPPA, PARADIGM)
1. Add heat 2. Diffuse heat 3. Cut out linkers
Evan Paull

51. HotLink
Genomic Perturbations Gene Activity
(Mutations, Methylation, Focal Copy Number) (Expression, RPPA, PARADIGM)
Linking Sub-Pathway
Evan Paull

52. HotLink “Double Diffusion” Overview

53. HotLink Causal Paths

54. Significance of Interlinking Network

55. Double diffusion better than single

56. Basal-LumA
HotLink
Basal LumA

57. Basal-LumA HotLink
MAP, PI3K, AKT Map
Basal LumA
TP53, RB1
MYC, FOXM1

58. AKT/PI3K
MAPK8, MAPK14 (p38-
alpha)
identified as mediators.

59. TP53, RB1
Basal LumA

60. MYC Neighborhood
Double feedback loop involving
TP53, RB1, CDK4, FOXM1, and MYC
Basal LumA

61. AKT signaling
Basal LumA

62. 



Ted Goldstein

63.  ’

Mutations
PARADIGM Signatures
Ted Goldstein

64.  ’

Mutations
APC and other Wnt
PARADIGM Signatures
Ted Goldstein
(Note: CRC figure below; soon for BRCA)

65.  ’

Mutations
PARADIGM Signatures
Ted Goldstein

66.  ’

Mutations
PARADIGM Signatures
TGFB Pathway mutations
Ted Goldstein
(Note: CRC figure below; soon for BRCA)

67.  ’

Mutations
PARADIGM Signatures
PIK3CA, RTK pathway, KRAS
Ted Goldstein
(Note: CRC figure below; soon for BRCA)

68.  ’

Mutations
PARADIGM Signatures
Evidence for
AHNAK2 acting
PI3KCA-like?
(Note: CRC figure below; soon for BRCA) Ted Goldstein

70. UCSC Integrative Genomics Group
Please See Posters!
Sam Ng Dan Carlin Evan Paull
Marcos Woehrmann
Ted Golstein
James Durbin Artem Sokolov Yulia Newton
Chris Szeto
Chris Wong

71. David Haussler
Buck Institute for Aging Chris Benz,
• Christina Yau
• Sean Mooney
UCSC Cancer Genomics
Jing Zhu • Janita Thusberg
• Kyle Ellrott
Collaborators
• Brian Craft
• Chris Wilks • Joe Gray, LBL
• Amie Radenbaugh • Laura Heiser, LBL
• Mia Grifford • Eric Collisson, UCSF
• Sofie Salama • Nuria Lopez-Bigas, UPF
• Steve Benz • Abel Gonzalez, UPF
Broad Institute
Funding Agencies
UCSC Genome Browser Staff • Gaddy Getz
• Mark Diekins • NCI/NIH
• Mike Noble
• SU2C
• Melissa Cline • Daniel DeCara
• NHGRI
• Jorge Garcia • AACR
• Erich Weiler • UCSF Comprehensive Cancer Center
• QB3

72. UCSC Cancer Browser
genome-cancer.ucsc.edu
Jing Zhu

73. supplemental

74. 325 Genes, 2213 Interactions

75. “Backbone” of 43 genes, 90 connections

76. “Backbone” of 43 genes, 90 connections
Master regulators: Cell-Cell Signaling, AKT1, Cyclin D1

77. “Backbone” of 43 genes, 90 connections
Major PARADIGM hubs included: MYC, FOXM1, FOXA1, HIF1A/ARNT

78. “Backbone” of 43 genes, 90 connections
Signaling through beta-catenin explains MYC activity in basals:
-deletions in CDKN2A de-repress CTNNB1 in basals or
-lower expression of Cyclin D1 de-repress CTNNB1

79. Top 20 basal master controllers
Yulia Newton

80. RNAi vs. Master Controller (after recurring
runs)
Basal vs Luminal RNAi Growth
AKT2
RPS6KA3
- p90 S6 kinase
PDPK1
AKT1
High-scoring after
Iterative runs.
RAF1
Basals differentially
RPS6KA3
Sensitive to RNAi
Inhibitors available.
Master Controller Score
Yulia Newton