Computational Methods for Systems and Synthetic Biology - Part IVContents

(1)

Computational Methods for Systems and Synthetic Biology - Part IV

 Jan 20: Introduction to logical modelling + MAPK network

 Jan 27: Blood cell specification and differentiation + GINsim tutorial

 Feb 03: Stochastic extension (G. Stoll and L. Calzone)

 Feb 10: From network structure to dynamics (E. Remy)

Paris, January 20th, 2015

Denis Thieffry ([email protected])

(2)

Boolean networks - Stuart Kauffman (1969)

The Boolean vector x represents the state of the system

Random connections, nodes with predefined degree

Canalizing Boolean functions Focus on asymptotic behaviour

Two types of attractors: stable states and (simple) cycles

Deterministic behaviour

(only one possible following state)

x _t ₊₁ = B( x _t )

(3)

Kinetic logic - René Thomas (1973)

X i (image or logical function) specifies whether gene i is currently transcribed x i (logical variable) denotes the presence (above a threshold of the functional

product of gene i

X = B( x )

t X i

Gene i switched ON Gene i switched OFF

1 0

x i

1 0

Delay d OFF

Delay d ON

Asynchronous updating

(4)

Logical variations: Model definition

Formalism Description Authors

Boolean equations Use of Boolean operators NOT, AND, and OR

Kauffman Threshold models Boolean rules derived directly from the

graph (e.g. sums of positive/negative inputs compared to thresholds)

Borhnoldt Li

Bipartite graph Introduction of AND nodes

(regulations converging onto a

components are combined with OR)

Klamt, Saez- Rodriguez Regulatory graph

+ Boolean functions

The regulatory graph constraints the

definition of the logical function Demongeot Goles

Irons Regulatory graph

+ multilevel functions

The regulatory graph constraints the definition of the logical function

Notion of logical parameters

Thomas Snoussi Fuzzy logic A third symbolic value enables logical

computation with unknown levels Lauffenburger

Sorger

(5)

Logical variations: updating schemes

Updating Description References

Synchronous All components are updated simultaneously (deterministic paths)

Kauffman Semi-

synchronous

Use of dummy nodes to delay defined components (deterministic paths)

Albert, Chaves, Irons

Bloc

synchronous

Components from a same bloc are updated

synchronously following rank (deterministic paths)

Goles,

Demongeot Fully

asynchronous

All enabled single transitions are considered (non deterministic transition graph)

Thomas Time delays Association of continuous clocks with components Thomas,

Bockmayr Mixed Synchronous or asynchronous priority classes

(non deterministic transition graph)

Fauré et al

(2006, 2009)

Stochastic Probabilities on transitions / Markov chains Stoll et al (2012)

(6)

Assets of logical modelling

• Exploitation of heterogenous, incomplete and/or qualitative data

• Versatility (e.g. consideration of different levels of abstraction)

• Bottom up approach (easy composition)

• Rigorous formal framework

• Scaling up potential, e.g. taking advantage of reduction methods

• Straightforward simulation of perturbations (KO, KI, etc.) => predictive power

• Powerful simulation and analysis tools

• SBML Qual exchange format - http://arxiv.org/abs/1309.1910

(7)

Logical modelling software tools

Software Availability References

Cell Collective (ChemChain)

Web based platform Helikar et al (2012)

CellNOpt R package

MATLAB package Cytoscape plugin

Klamt et al (2009) Terfve et al (2012)

GINsim Java package Chaouiya et al (2003, 2013)

MaBoSS C++ package Stoll et al (2012)

SQUAD Java package Di Cara et al (2007)

booleannet Python package Albert et al (2008)

BoolNet R package

(RBN generation)

Müssel et al (2010)

DDlab C/C++ package Wuensche (2011)

(8)

p53 regulation

Hasty et al (2001)

(9)

A simple model for the p53-Mdm2 network

Regulatory Graph

[2]

DNAdam => 1 IFF DNAdam & !p53

DNAdam

p53

0 0 1 0

p53

Decision tree

(10)

A simple model for the p53-Mdm2 network

Regulatory Graph

[2]

DNAdam => 1 IFF DNAdam & !p53

DNAdam

p53

0 1

Decision diagram

(11)

A simple model for the p53-Mdm2 network

Regulatory Graph

[2]

Logical rules

Components Target levels Boolean rules

p53 2 !Mdm2nuc

Mdm2cyt 1 p53:2

Mdm2nuc 1 Mdm2cyt | !p53 & !DNAdam

DNAdam 1 DNAdam & !p53

!, | and & stand for the Boolean operators NOT , OR and AND, respectively.

Default: other conditions leads each component => 0

(12)

A simple model for the p53-Mdm2 network

Regulatory Graph

[2]

Components Target levels Parameters

p53 2 K ^p53 {}

Mdm2cyt 1 K ^Mdm2cyt { ^p53 }

Mdm2nuc 1 K ^Mdm2nuc {}

K ^Mdm2nuc { ^Mdm2cyt } K ^Mdm2nuc { ^Mdm2cy,p53 } K ^Mdm2nuc { DNAdam, Mdm2cyt } K ^Mdm2nuc { DNAdam, Mdm2cyt, p53 }

DNAdam 1 K DNAdam { DNAdam }

Logical parameters

Default: other conditions leads each component => 0

(13)

Full asynchronous simulation:

state transition graph (STG)

[p53,Mdm2cyt, Mdm2nuc, DNAdam]

Stable (resting) state

with only Mdm2nuc ON

(14)

Full synchronous simulation:

state transition graph (STG)

[p53,Mdm2cyt, Mdm2nuc, DNAdam]

Stable (resting) state with only Mdm2nuc ON

Terminal cycle Terminal cycle

(15)

Full asynchronous simulation:

graph of strongly connected components (SCC)

(16)

Full asynchronous simulation:

SCCG vs HTG

SCCG HTG

The HTG is generated using a modified version of Tarjan's algorithm

(Bérenguier et al, 2013).

(17)

GINsim (Gene Interaction Networks simulation)

analysis toolbox core simulator

GINML parser user interface

graph analysis graph

editor simulation

State transition graph

Regulatory graph

Available at

http://ginsim.org

Aurélien NALDI Pedro MONTEIRO Claudine CHAOUIYA

Chaouiya et al (2012) Meth Mol Biol 804: 463-79

Duncan BERENGUIER

Lionel SPINELLI

(18)

Development of dynamical analysis tools

 Decision diagrams

• Identification of attractors

• Analysis of regulatory circuits

• Model reduction

• State transition graph compression

 Model checking

• Verification of dynamical properties (temporal logic)

 Priority classes

• Mixed a/synchronous simulations

 Petri nets

• Standard Petri nets

• Coloured Petri nets

(19)

Efficient identification of stable states

A ^B

C

1 0

A

K C =1 IFF !A

1 0

A

C

K B =1 IFF A & !C K A = 1 IFF A

0 1

A

=> 2 stable states : 001 et 110

(20)

=> 2 stable states : 001 et 110

Efficient identification of stable states

A ^B

C

0 1

0 A

C C

K

C

1 0 1 0

A

B B

C C

stable

unstable

K

B

*

0 1 0 0

A

C B B

C

Stability condition

*

Naldi, Chaouiya & Thieffry (2007) LNCS 4695: 233-47.

K

A

1 1

A

stable

(21)

Roles of regulatory circuits (R Thomas)

Characteristics Positive circuits Negative circuits Number of negative

interactions Even Odd

Dynamical property

Maximal level

Bottom level

Biological property Differentiation Homeostasis

Examples cI cro Cro

(22)

Regulatory circuits: dynamics in isolation

stable states

attracting cycle

A

B C

D

Positive circuit

A

B C

D

Negative circuit

Remy et al (2003) Bioinformatics 10: ii172-8

(23)

Regulatory circuits & Thomas' rules

 A positive regulatory circuit is necessary to generate multiple stable states or attractors

 A negative regulatory circuit is necessary to generate sustained oscillatory behaviour

 Thomas R (1988). Springer Series in Synergics 9: 180-93.

Mathematical theorems and demonstrations:

 In the differential framework:

Thomas (+, 1994), Plathe et al. (±, 1995), Snoussi (±, 1998), Gouzé (±, 1998), Cinquin & Demongeot (+, 2002),

Soulé (+, 2003).

 In the discrete framework:

Aracena et al. (+, 2001), Remy et al. (±, 2005),

Richard & Comet (+, 2005).

(24)

A

B C

D

A = a

B = a c C = b

D = c v

abc BCD 000! 110 001! 011 010! 100 011! 001 100! 010 101! 011 110! 000 111! 001

Regulatory circuit functionality

Circuit properties depends on the effect of A on B

 In the presence of A: → only one stable state with {A,B,C,D}= 1011

 In the absence of A: → two stable states 0100 and 0011

 The positive circuit is thus functional only in the absence of A

(25)

Take home messages

• Qualitative versus (often illusory) quantitative aspects

• Dynamical roles of feedback circuits

• Flexibility of the generalised logical formalism

• Beware of the artefacts of the synchronous updating method

(26)

Logical modelling

of MAPK signalling network

(27)

JNK1, JNK2, JNK3 MLK2/3,

MEKK1-4, ASK1, TAK1, TAOK1/2

p38α, p38β, p38γ, p38σ A-RAF, B-RAF,

C-RAF

MEK1, MEK2

ERK1, ERK2

MAPKKK

MAPKK

MAPK

MEK3, MEK6 MEK4, MEK7

MLK2/3, MEKK1-4, ASK1,

TAK1, TAOK1/2

Proliferation, Apoptosis, Differentiation, etc.

Growth factors, Cytokines, Stresses, etc.

Simplistic view of MAPK pathways

(28)

Specificity factors

• Different stimuli

• Protein isoforms

• Sub-cellular localisation

• Scaffold proteins

• Phosphatases

• ^Feedbacks

• Cross-talks

Yao and Seger, BioFactors, 2009

(29)

Biological questions

Generic scope

Study the role(s) of MAPK signalling deregulations in cancer cell fate decision

(Im)balance between

Proliferation

Growth arrest / Apoptosis

Specific aims

• Identification of key players for the transduction of proliferative signals in bladder cancers

• Understand the mechanisms governing varying MAPK

activity in urinary bladder cancer subtypes

(30)

Computational systems biology approach

Focus on the structure and dynamics of the entire network, with the help computational methods

Integration of relevant information into a

detailed reaction map

Dynamical modelling to study the behaviour of MAPK network in cancer

Static pathway analysis to identify key proliferative

components of MAPK network

(31)

The MAPK reaction map

• CellDesigner software (www.celldesigner.org)

• Literature-derived information

• Emphasis on specificity factors

• Generic map: several human/mouse cell types

Integration of relevant information into a detailed reaction map

Luca GRIECO

now at UCL, UK

(32)

The MAPK reaction map

- 248 distinct components (proteins, complexes, genes, ...) - 176 reactions

∼ 200 articles

Each component and reaction is annotated regarding

information sources and modelling choices

(33)

External stimuli and phenotypes

RTK GPCR TNFR IL1R TGFβR

Apoptosis Proliferation Growth arrest

(34)

Sub-cellular compartments

Plasma membrane

Endoplasmic

reticulum Mitochondria

Early endosomes Golgi

apparatus Late

endosomes

Nucleus

Genes

Phenotypes

(35)

Example: ERK activation

(36)

Clickable map - Atlas Of Cancer Signalling Networks

Using BiNoM plugin of Cytoscape

https://acsn.curie.fr/navicell/maps/acsn/master/index.html

(37)

Dynamical modelling of MAPK network

• Focus on urinary bladder cancer

• Collaboration with François Radvanyi’s group (Institut Curie) Integration of relevant

information into a

detailed reaction map

Dynamical modelling to study the behaviour of MAPK network in cancer

Static pathway analysis to identify key proliferative

components of MAPK network

(38)

Dynamical modelling of MAPK pathways

Authors Formalism Dimension Description

Huang & Ferrel (1996) ODE 18 Ultrasensitivity of MAPK cascades depending on their particular structure:

double phosphorylation and double de-phosphorylation mechanisms.

Kholodenko (2001) ODE 8 Sustained oscillations of MAPK phosphorylation level, induced by negative feedback loop and cascade ultrasensitivity.

Levchenko et al (2000) ODE 27 Effects of scaffold proteins on MAPK activation.

Brightman & Fell (2000) ODE 29 Transient versus sustained ERK activation following growth signals.

Schoeberl et al (2002) ODE 94 Dynamics of EGF-dependent ERK activation.

Hatakeyama et al (2003) ODE 33 Interplay between PI3K-AKT and RAF-ERK pathways in EGFR signalling network.

Markevich et al (2004) ODE 15 Bistable behaviour of MAPK cascades due to multistep phosphorylation/

dephosphorylation cycles.

Hornberg et al (2005) ODE 103 Identification of reactions controlling amplitude, duration and integrated output of EGF-dependent ERK signalling network.

Sasagawa et al (2005) ODE 22 Transient versus sustained ERK activation following growth signals.

Legewie et al (2007) ODE 21 Bistability of ERK activation due to MEK-ERK positive feedback circuit.

Smolen et al (2007) ODE 26 Mechanisms contributing to the bistable behaviour of MAPK cascades.

Samaga et al (2009) Boolean model 104 Topological properties and qualitative behaviour of the EGFR signalling network.

Sturm et al (2010) ODE 14 Negative feedback amplifier properties of ERK cascade.

Handorf & Klipp (2011) Boolean model 107 Boolean model of cross-talk between WNT and ERK pathways, semi- automatically derived from a public reaction database.

Bagheri et al (2011) ODE 4 Predictions of MEK inhibition-based treatment effects on tumour cell proliferation.

Finch et al (2012) ODE 6 Phosphatase-mediated cross-talk between p38 and ERK cascades.

Poltz & Naumann (2012) Boolean model 96 DNA damage response network in human epithelial tumours; identification of target molecules for DNA damaging therapies.

Chen et al (2012) ODE 478 Model of immediate-early signalling involving ErbB receptors, along with

MAPK and PI3K pathways, trained against time course experimental data.

(39)

Bladder cancer

Non-invasive

Invasive

(40)

Focus

Mechanisms governing the behaviour of urinary bladder cancer subtypes, in response to selected stimuli

Bladder cancer deregulations

• EGFR over-expression

• FGFR3 activating mutation

• Both receptors activate MAPKs

Invasive (>T1) &

high proliferation rate Ta & less aggressive

Aims

1. Recapitulate this differential behaviour with a dynamical model

2. Decipher the underlying mechanisms

(41)

The MAPK logical model

Influence graph, 53 components, 552 circuits (complexes => implicit)

(42)

re4 re3; re5

re5 re6

CellDesigner

The MAPK logical model

(43)

WILD TYPE

p53=1 iff (ATM=1 and p38=1)

or ((ATM=1 or p38=1) and MDM2=0) p53=0 otherwise

The MAPK logical model

(44)

WILD TYPE

p53=1 iff (ATM=1 and p38=1)

or ((ATM=1 or p38=1) and MDM2=0) p53=0 otherwise

PERTURBATIONS

- p53 loss-of-function: p53=0 always - p53 gain-of-function: p53=1 always

The MAPK logical model

(45)

2 ⁵³ states

The MAPK logical model

(46)

Coping with the exponential growth of logical state transition graphs

 Focus on attractors and their reachability

 Model reduction (based on user specifications)

 Compaction of state transition graphs

 Temporisation (e.g. priorities, delays, etc.)

 Model checking

 Delineation of the roles of regulatory circuits/modules

(47)

Model reduction (GINsim)

X T

T R3

R1

R3 R1

R2

Keep the detailed model Reduction before analysis

=> New rules for targets of hidden nodes

Choice of reduction

Dynamical consistency - No circuit deletion

- Same stable states

- Reachability may change

Naldi et al. (2011)

Theoretical Computer Science 412: 2207-18

(48)

Reduction 1 Reduction 2 Reduction 3 Inputs EGFR_stimulus, FGFR3_stimulus, TGFBR_stimulus,

DNA_damage

EGFR_stimulus, FGFR3_stimulus, TGFBR_stimulus, DNA_damage

Phenotypes Proliferation, Apoptosis, Growth_Arrest Proliferation, Apoptosis, Growth_Arrest Proliferation, Apoptosis, Growth_Arrest Selected

observables

EGFR, FGFR3, p53, p14, PI3K, AKT, PTEN, ERK

EGFR, FGFR3, RAF, RAS, ERK, AKT, p53, p21

JNK, p38,

GADD45, ERK, RAS

Auto-

regulated components

FRS2, MSK GRB2, PI3K, p38 GRB2, PLCG, PI3K, MDM2

Three reductions of MAPK models

Each reduction preserves the input and phenotype components.

Additional components were kept depending on the simulations performed.

Apparition of auto-regulations impede further component reduction.

Conservation (compression) of regulatory circuit ensure the preservation of the main dynamical properties.

MAPK model reduction

(49)

MAPK reduced model (version 1)

17 components (including 4 inputs and 3 outputs), 128 circuits

Functional circuits: 1 positive, 5 negative, 1 dual

(50)

Asynchronous simulation for p53 KO Hierarchical State Transition Graph

HTG dimension: 21 nodes STG dimension: 637 nodes

Init. cond.: FGFR3_stim = 1

ERK+

Spry+

GRB2-

p38-

JNK-

PI3K+

(51)

Simulations of EGFR vs FGFR3 activating mutations

(52)

Simulations of documented perturbations

(53)

Attractors for the MAPK model

Perturbations Inputs_set_to_1 Attractor type Apoptosis Growth_Arrest Proliferation ERK p53 EGFR FGFR3 FRS2 PI3K AKT MSK p14 PTEN none EGFR_stimulus Cyclic

(#640) * * * 0 * * 0 0 1 * * * *

none EGFR_stimulus Cyclic

(#640) * * * 1 0 * 0 0 1 * 1 * *

none FGFR3_stimulus Cyclic

(#1344) * * * 0 * 0 * * 1 * * * *

none FGFR3_stimulus Cyclic

(#1344) * * * 1 0 0 * * 1 * 1 * *

none FGFR3_stimulus Cyclic (#1344)

* * * 1 0 1 0 0 1 * 1 * *

EGFR gain none Stable state 0 0 1 1 0 1 0 0 1 1 1 1 0

EGFR gain none Stable state 1 1 0 0 1 1 0 0 1 0 1 1 1

FGFR3 gain none

Stable state 0 0 1 1 0 0 1 0 1 1 1 1 0

FGFR3 gain none Stable state 0 0 0 1 0 0 1 0 0 0 1 0 0

FGFR3 gain none

Cyclic (#2) 1 1 0 0 1 0 1 * 1 0 1 1 1

EGFR gain

p53 loss none Stable state 0 0 1 1 0 1 0 0 1 1 1 1 0

FGFR3 gain

EGFR gain DNA_damage Stable state 1 1 0 0 1 1 0 0 1 0 1 1 1

FGFR3 gain DNA_damage Cyclic (#2) 1 1 0 0 1 0 1 * 1 0 1 1 1

EGFR gain

FGFR3 gain

EGFR gain PI3K gain

AKT gain none Stable state 0 0 1 1 0 1 0 0 1 1 1 1 0

EGFR gain PI3K gain

FGFR3 gain PI3K gain

AKT gain none

Cyclic (#2) 0 0 0 0 1 0 1 * 1 1 1 1 1

EGFR gain TGFBR_stimulus Stable state 1 1 0 0 1 1 0 0 1 0 1 1 1

FGFR3 gain TGFBR_stimulus Stable state 1 1 0 0 1 0 1 0 1 0 1 1 1

EGFR gain

PTEN loss none Stable state 0 0 1 1 0 1 0 0 1 1 1 1 0

EGFR gain

FGFR3 gain

PTEN loss none

Stable state 0 0 1 1 0 0 1 0 1 1 1 1 0

FGFR3 gain

PTEN loss none

Cyclic (#2) 0 0 0 0 1 0 1 * 1 1 1 1 0

(54)

Coherence of simulations with published data

Biological data Model behaviour

RAF or RAS over-expressions can lead to constitutive activation of ERK.

In absence of inputs, constitutive activity of RAF or RAS can lead to permanent ERK activation, associated with proliferation.

HSP90-inhibitor disrupts RAF, AKT and EGFR,

leading to successful cancer treatment. Concomitant RAF, AKT, EGFR deletions abrogate the proliferative stable states, in the case of EGFR over-expression and in the case of FGFR3 activating mutation.

Patients with p53-altered/p21-negative tumors demonstrated a higher rate of recurrence and worse survival compared with those with p53- altered/p21-positive tumors.

Following either EGFR over-expression or FGFR3 activating mutation, concomitant p21 and p53 loss-of-functions correspond to a phenotype characterised by apoptosis escape, with the possibility to attain

proliferation. Association of p53 loss-of-function and p21 gain-of- function leads to growth arrest attractors, without proliferation.

p38 and JNK play important roles in stress

responses, such as cell cycle arrest and apoptosis. In presence of either DNA_damage or TGFBR_stimulus, growth arrest/

apoptosis stable states are all lost in the p38/JNK-deleted model.

p38 and JNK have been shown to induce apoptotic cell death.

When p38/JNK are constitutively active, apoptotic attractors are obtained in the absence of other stimuli.

p38 plays its tumour suppressive role by promoting

apoptosis and inhibiting cell cycle progression. Under JNK constitutive activation, p38 loss-of-function determines loss of apoptotic attractors obtained in r26.

JNK may contribute to the apoptotic elimination of transformed cells by promoting apoptosis.

Under p38 constitutive activation and JNK loss-of-function, apoptotic attractors are lost.

Epigenetic gene silencing of GADD45 family members has been frequently observed in several types of human cancers.

In presence of DNA_damage), Growth_Arrest and Apoptosis components permanently oscillate when GADD45 is silenced,

suggesting less propensity to cell death. Apoptotic stable states are still reached in presence of TGFBR_stimulus

ERK increases transcription of the cyclin genes and facilitates the formation of active Cyc/CDK

complexes, leading to cell proliferation.

ERK gain-of-function always leads to proliferative attractors, in the absence of other stimuli.

ERK disrupts the anti-proliferative effects of TGFβ. TGFBR_stimulus leads to an apoptotic stable state, but coupling of TGFBR_stimulus with ERK gain-of-function leads to growth arrest.

JNK might reduce RAS-dependent tumour

formation by inhibiting proliferation and promoting apoptosis.

In absence of other stimuli, JNK constitutive activation completely abrogates RAS-dependent proliferation following RAS over-

expression. Instead, apoptotic attractors are always reached.

(55)

Using the model to analyse feedback mechanisms

(56)

Simulations of the disruption of GRB2 vs Sprouty feedback on FRS2 under FGFR3 gain-of-function

X

(57)

Apoptosis Growth_Arrest Proliferation ERK p53 EGFR FGFR3 FRS2 PI3K AKT MSK p14 PTEN

0 0 1 1 0 0 1 0 1 1 1 1 0

0 0 0 1 0 0 1 0 0 0 1 0 0

FGFR3 stimulation and multistability

(58)

Comparison with experimental data

PI3K activation tentatively influences the switch between proliferation and growth arrest following FGFR3 stimulus:

In FGFR3-mutated bladder cancer cell lines, PI3K activation (in contrast with MAPKs) is determinant for proliferation

(unpublished data from Radvanyi’s group, Institut Curie).

FGFR3_stimulus (alone)

Proliferation

Growth Arrest

ERK = 1

p38 = 0

JNK = 0

PI3K = 1

ERK = 1

p38 = 0

JNK = 0

PI3K = 0

(59)

Outlook

‣ Qualitative recapitulation of known effects of MAPK network on cancer cell fate decision, following specific stimuli

‣ Insights into the role of MAPK network and of specific components in different bladder cancer types

‣ Novel hypotheses concerning the mechanisms underlying the different effects of EGFR/FGFR3 deregulations

- Feedbacks via Sprouty

- PI3K switch following FGFR3 stimulus

(60)

Prospects

‣ Use of the molecular map for the analysis of (functional) genomic data (clinical samples)

‣ Automatic derivation of the logical model from the reaction map e.g. using a rule based modelling approach

(collaboration with Jérȏme Feret - ENS Paris)

‣ Use of the logical model to predict synergies between drugs

‣ Refinement of the logical model using novel data

(additional components and feedbacks - e.g. phophatases, sprouty and other protein variants -, use of multilevel components, etc.)

‣ Towards more quantitative analyses (e.g. using a probabilistic framework)

‣ Module for more comprehensive cell fate models

‣ Validation of model predictions (role of Sprouty)

(61)

• Abou-Jaoudé W, Ouattara D, Kaufman M (2009). From structure to dynamics:

frequency tuning in the p53-Mdm2 network I. Logical approach. Journal of Theoretical Biology 258: 561–77.

• Abou-Jaoudé W, Monteiro PT, Naldi A, Grandclaudon M, Soumelis V, Chaouiya C, Thieffry D (in press). Dynamical analysis of logical signalling networks through

reduction methods and model-checking. Frontiers in Bioengineering and Biotechnology.

• Bérenguier D, Chaouiya C, Monteiro PT, Naldi A, Remy E, Thieffry D, Tichit L (2013).

Dynamical modeling and analysis of large cellular regulatory networks. Chaos 23:

025114.

• Grieco L, Calzone L, Bernard-Pierrot I, Radvanyi F, Kahn-Perlès B, Thieffry D (2013).

Integrative modelling of the influence of MAPK network on cancer cell fate decision.

PLoS Computational Biology 9: e1003286.

• Naldi A, Carneiro J, Chaouiya C, Thieffry D (2010). Diversity and plasticity of Th cell types predicted from regulatory network modelling. PLoS Computational Biology 6:

e1000912.

• Thieffry D (2007). Dynamical roles of biological regulatory circuits. Briefings in Bioinformatics 8: 220-5.

• Thomas R (1991). Regulatory networks seen as asynchronous automata: a logical

description. Journal of theoretical Biology 153: 1-23.

(62)

Collaborations & supports

★ ENS (Paris)

•

Wassim Abou-Jaoudé

•

Samuel Collombet

•

Jérome Feret

•

Anna Niarakis

•

Morgane Thomas-Chollier

★ Institut Curie (Paris)

•

Emmanuel Barillot

•

Isabelle Bernard-Pierrot

•

Eric Bonnet

•

Laurence Calzone

•

Philippe Hupé

•

Francois Radvanyi

•

Vassili Soumelis

•

Maxime Touzot

•

Andrei Zinovyev

★ TAGC (Marseille)

•

Luca Grieco (=> UCL, London)

•

Brigitte Kahn-Perlès

•

Aurélien Naldi (=> UNIL, Lausanne)

•

Jacques van Helden

★ IML (Marseille)

•

Duncan Berenguier

•

Elisabeth Rémy

★ IGC (Lisboa)

•

Claudine Chaouiya

•

Jorge Carneiro

•

Pedro Monteiro

Belgian Inter-university Attraction Pole

Bioinformatics and Modelling : from Genomes to Networks

(63)

Selected references

 Abou-Jaoudé W, Monteiro PT, Naldi A, Grandclaudon M, Soumelis V, Chaouiya C, Thieffry D (2015). Dynamical analysis of logical signalling networks through reduction methods and model-checking. Frontiers in Bioengineering and Biotechnology, in press.

 Bérenguier D, Chaouiya C, Monteiro PT, Naldi A, Remy E, Thieffry D, Tichit L (2013).

Dynamical modeling and analysis of large cellular regulatory networks. Chaos 23:

025114.

 Grieco L, Calzone L, Bernard-Pierrot I, Radvanyi F, Kahn-Perlès B, Thieffry D (2013).

Integrative modelling of the influence of MAPK network on cancer cell fate decision.

PLoS Computational Biology 9: e1003286.

 Naldi A, Thieffry D, Chaouiya C (2007). Decision diagrams for the representation and analysis of logical models of genetic networks. Lecture Notes in Bioinformatics 4695:

233-47.

 Naldi A, Remy E, Thieffry D, Chaouiya C (2011). Dynamically consistent reduction of logical regulatory graphs. Theoretical Computer Science 412: 2207-18.

 Naldi A, Carneiro J, Chaouiya C, Thieffry D (2010). Diversity and plasticity of Th cell types predicted from regulatory network modelling. PLoS Computational Biology 6:

e1000912.

 Thomas R (1991). Regulatory networks seen as asynchronous automata: a logical

description. Journal of theoretical Biology 153: 1-23.

Computational Methods for Systems and Synthetic Biology - Part IVContents