Bayesian network models of Biological signaling pathways

1
Bayesian network models of Biological signaling
pathways

• karensachs_at_stanford.edu

2
From Phospho-molecular profiling to Signaling
pathways
Cell1
Cell2
Cell3
Flow Measurments
Cell4
...
Cell600
Picture John Albeck
Signaling Pathways
High throughput data
3
Outline

• What are signaling pathways?
• What kind of data is available study them?
• How do we use Bayesian networks to learn their
  structure?

• Two extensions
• Markov neighborhood algorithm
• Bayesian network based cyclic networks (BBCs)

4
Outline

• What are signaling pathways?
• What kind of data is available study them?
• How do we use Bayesian networks to learn their
  structure?

• Two extensions
• Markov neighborhood algorithm
• Bayesian network based cyclic networks (BBCs)

5
Cells respond to their environment
6
Central Dogma
DNA
Modified Protein
Delivers instructions for specific gene
Ribosome Protein-production factory
Blueprint- instructions for production of all
proteins
7
Signaling Genetic pathways
A
B
C
Cell response
DNA
8
Outline

• What are signaling pathways?
• What kind of data is available study them?
• How do we use Bayesian networks to learn their
  structure?

• Two extensions
• Markov neighborhood algorithm
• Bayesian network based cyclic networks (BBCs)

9
Spectrum of Modeling Tools in Systems Biology
10
Bayesian Networks
Protein A
Protein B
Protein E
P(BAOn)
Protein C
Protein D

• Graph
• Node Measured level/activity of protein
• Edge Influence (dependency) between proteins
• Conditional probability distributions
• Each node has a conditional probability given its
  parents

10
11
How do we use Bayesian Networks to infer pathways?
The Technical Details
? Score candidate models
? Use a heuristic search to find high scoring
models
12
Protein data

• Western blot

13
Protein data

• Protein arrays

14
Protein data

• Mass Spectrometry

All of these lysate approaches give 1 measurement
per protein for 103-107 cells
15
Flow Cytometry Single Cell Analysis
Thousands of datapoints
16
Stimulations and perturbations
LFA-1
CD3
CD28
L A T
RAS
Cytohesin
PI3K
JAB-1
Zap70
10
Lck
PKC
PLCg
Akt
PIP3
Activators 1. a-CD3 2. a-CD28 3. ICAM-2
4. PMA 5. b2cAMP Inhibitors 6. G06976
7. AKT inh 8. Psitect 9. U0126 10.
LY294002
PIP2
PKA
Raf
MAPKKK
Mek1/2
MAPKKK
MEK3/6
Erk1/2
MEK4/7
p38
JNK
17
T-Lymphocyte Data

• Datasets
• of cells
• condition a
• condition b
• conditionn

• Primary human T-Cells
• 9 conditions
• (6 Specific interventions)

• 9 phosphoproteins, 2 phospolipids
• 600 cells per condition
• 5400 data-points

Omar Perez
18
Statistical Dependencies
Phospho A
Phospho B
19
Statistical Dependencies
Phospho A
Phospho B
Edges can be directed (primarily) due to the use
of interventions
20
Overview
Influence diagram of measured variables
Bayesian Network Analysis
21
Inferred Network
Phospho-Proteins

Phospho-Lipids

PKC
Perturbed in data

PKA
Raf
Plc?
Jnk
P38
Mek
PIP3
P44/42
Akt
PIP2
22
How well did we do?
Phospho-Proteins

Phospho-Lipids

PKC
Perturbed in data

PKA
Raf
Plc?
Jnk
P38
Mek
PIP3
P44/42
Akt
PIP2
Direct phosphorylation
23
Features of Approach

• Direct phosphorylation

Mek
Erk
Difficult to detect using other forms of
high-throughput data -Protein-protein
interaction data -Microarrays
24
How well did we do?
Phospho-Proteins

Phospho-Lipids

PKC
Perturbed in data

PKA
Raf
Plc?
Jnk
P38
Mek
PIP3
P44/42
Akt
PIP2
25
How well did we do?
Phospho-Proteins

Phospho-Lipids

PKC
Perturbed in data

PKA
Raf
Plc?
Jnk
P38
Mek
PIP3
P44/42
Akt
PIP2
Indirect Signaling
26
Indirect signaling

• Indirect signaling

Indirect connections can be found even when the
intermediate molecule(s) are not measured

• Dismissing edges

27
Indirect signaling - Complex example

• Is this a mistake?
• The real picture
• Phoso-protein specific
• More than one pathway of influence

28
How well did we do?
Phospho-Proteins

Phospho-Lipids

PKC
Perturbed in data

PKA
Raf
Plc?
Jnk
P38
Mek
PIP3
P44/42

• 15/17 Classic

Akt
PIP2
29
Signaling pathway reconstruction
Phospho-Proteins

Phospho-Lipids

PKC
Perturbed in data

PKA
Raf
Plc?
Jnk
P38
Mek
PIP3
Erk

• 15/17 Classic
• 17/17 Reported
• 3 Missed

Akt
PIP2
Sachs et al 2005
30
Caveats

• Inhibitor specificity
• Binding site similar across proteins
• Reagent availability and specificity
• Data quality
• These are issues in many biological apps!

I think Ill bind here
31
Outline

• What are signaling pathways?
• What kind of data is available study them?
• How do we use Bayesian networks to learn their
  structure?

• Two extensions
• Markov neighborhood algorithm
• Bayesian network based cyclic networks (BBCs)

32
Markov Neighborhood Algorithm
33
Building larger networks

• 12 color capability ? Model 50-100 variables
• 4 color capability ? Model 12 variables

80 proteins involved in MAPK signaling (11- at
the cutting edge- is NOT enough!)
33
34
Measured subsets Incomplete dataset (Missing
data)

• Insufficient information for standard approaches
  (will perform poorly)
• Use a set of biologically motivated assumptions
  to constrain search..
• And to reduce the number of experiments

34
35
Constraining the search

• Using Markov neighborhoods
• (for each variable)

• Plus potential perturbation parents

Identify candidate parents
35
36

Approach overview
Bayesian Network Analysis (Constrained search)
37

Neighborhood reduction
4?11
C
4 color capability
Conditional independencies in the substructure?
A?B?C
37
38
Accurate Reproduction of Model 15 experiments,
4-colors
PKC
PKA
Raf
Plc?
Jnk
P38
Mek
PIP3
Erk
Akt
PIP2
39

Active learning approach
39
40
Outline

• What are signaling pathways?
• What kind of data is available study them?
• How do we use Bayesian networks to learn their
  structure?

• Two extensions
• Markov neighborhood algorithm
• Bayesian network based cyclic networks (BBCs)

41
Learning cyclic structures with Bayesian networks

• Biological networks contain many loops
• Bayesian networks are constrained to be acyclic
• So

42
Overcoming acyclicity

• Signaling pathways contain many cycles
• Bayesian networks are constrained to be acyclic
• How can we accurately model pathways with cycles?

GRB2/SOS
Ras
Raf
MEK
Develop a new, Bayesian network derived algorithm
that models cycles ?
Erk
43
Bayesian Network Based Cyclic Networks (BBNs)

• I. Break loops with molecule inhibitors
• II. Use BN to learn the structure (now not
  cyclic!)
• III. Close loops

GRB2/SOS
Ras
Raf
Mek inhibitor
Solomon Itani
MEK
Erk
44
Bayesian Network Based Cyclic Networks (BBNs)

• I. Break loops with molecule inhibitors
• Detect loops P(A)A P(A)
• II. Use BN to learn the structure (now not
  cyclic!)
• III. Close loops
• P(BPa(B)) A P(BPa(B))
• A?B

GRB2/SOS
Ras
Raf
MEK
Erk
45
Future work

• Larger network from overlapping sets (Markov
  neighborhood)
• Dynamic models over time
• Differences in signaling (sub-populations,
  treatment conditions, cell types, disease states)

46
Acknowledgements
Garry Nolan
Dana Peer
Doug Lauffenburger
Omar Perez
Dennis Mitchell
Funding LLS post doctoral fellowship
Shigeru Okumura
Mesrob Ohannessian
Solomon Itani
46
47
Extra slides
48
Mathematical Intuition
B
C
A
C is independent of A given B.
A
B
C independent of A given B and D
C
D

1. No need to introduce time!!!
2. When loops are broken, the result is a BN!!!

49
Prediction Erk?Akt

• Erk1/2 unperturbed
• Erk ? Akt not well established in literature
• Predictions
• Erk1/2 influences Akt
• While correlated, Erk1/2 does not influence PKA

PKC
PKA
Raf
Mek
Erk1/2
Akt
50
Validation

• SiRNA on Erk1/Erk2
• Select transfected cells
• Measure Akt and PKA

P9.4e-5
P0.28