Franck Molina Sophia Antipolis juillet 2008 FRE3009 CNRS / BIO-RAD Modélisation et ingénierie des systèmes complexes biologiques pour le diagnostic Biological processes modelling based on elementary actions and synthetic biology
Franck MolinaSophia Antipolis juillet 2008
FRE3009 CNRS / BIO-RAD
Modélisation et ingénierie des systèmes complexes biologiques pour le diagnostic
Biological processes modelling based on elementary actionsand synthetic biology
Complex System Modeling and Engineering for Diagnosis
Alliance betweenBIO-RAD and CNRS
Diagnosis, biotechnologies Discovery, Scientific research
SysDiag est dédiée à la recherche de diagnostic en santé humaine(Cancer, Alzheimer, cardiovasculaire, diabète, etc.)
Comprendre les bases de maladies multifactorielles (Alzheimer, Prion, Hépatite, Cancer, Cardiovasculaire, Diabète…)Et d’identifier les biomarqueurs associés à ces conditions pathologiques
SysDiag a pour objectif d’accélérer le transfert des découvertes de la recherche scientifique vers des applications mises sur le marché.
Modélisation et ingénierie des systèmes complexes biologiques pour le diagnostic
Crossing disciplines for innovation in Diagnosis and Life Sciences
Combining experimental biology and complex systems modelling approaches
70% experimentalists and 30% theoreticians
“Task-team” organization
The SysDiag model
Complex system modeling and engineering for diagnostic
SiliCell,
Immunology / BiotechnologiesAntibodies, proteomics, mass specSELDI, Prot-Prot Interactions, Bio-plex, Proteon, peptide and protein arrays, mRT-PCRHT antibody facilities.
Modeling and computing approachesMolecular modelingInformation systemsComplex systems modelingBiostatistics
Crossing disciplines for innovation in Diagnosis and Life Sciences
Complex pathology diagnosis : Causalities and biomarkers
New approaches for tomorrow’sdiagnosis
Clinicians
samplesExpertise
Path
olog
ies
of in
tere
st
PublicationsPatents
SoftwaresAlgorithms
TechnologiesMethodologies
BiomarkersDiagnostic probes
Diagnostics processesMolecular mecanisms
PlatformChemistryInteraction
Clinical Proteomic
PlatformBioinformaticsBiostatistics
PlatformAntibody Dev Pf
Bioinformati
csmolec
ularBiology
Immunology
Chemist
ry
biochem
istry
Mathemati
cs
SysDiag Platforms
SPR : ProteOn
CIP Peptides Chemistry-Interactions-ProteomicsHT Peptide synthesis , SPOT technique
Bioplex, Proteon SPR, protein arrays
2Dproteomics,MALDI, SELDI.
HT Antibody plateformHT mAb generation, screening and characterization
Mimétisme moléculaire:Localisation d’épitopes à partir d’approches phage display : MIMOP Prédiction de peptides antigéniques pour des immunisations ciblées : PEPOPBioinformatics 2006, BMC bioinformatics 2008
Amm8
Reactivity of "discontinuous" peptides designed by PEPOP from the 3D model of Amm8
Amm8
Lack of reactivity of the 15-mer overlapping peptides with anti-Amm8 antibody
Bioinformatics and biostatisticsMolecular modeling, epitope analysis, antigen prediction
Patient
followup
THERANOSTIC
Drug & Diag Discovery
DIA
GD
ISC
OVE
RY
DR
UG
DIS
CO
VER
YP
RA
CTI
CIA
NS
Questions Pathology characteristicsTherapeutic decision
Therapeuticproblem
Targetidentification
Activecompound
Clinicalstudies
Diagnosisproblem
Biomarkeridentification
Bindingprobes
Clinicalstudies
TEST DIAG
DRUG
treatment
MonitoringDiagnosis
AMA
AMA
2-4 years 1-2 years 1 year
3-6 years 6-8 years 1 year10 – 14 years
4 – 7 years
'Application for Market Authorization'
Clinical samples
Biomarkeridentification
Experimentalanalyses
Biomarkeridentification
Signal analysesBiostatistics
MolecularMechanism
understanding
Validation
SetOf
Biomarkers
Diagnostic Assay design
Specific probes
Research for early diagnosis and follow up of complex diseases
Qualityassessment
Systems BiologyComplex system modeling
HT mAb platfomand
Multiplexed assaysProfiling and omics
Biostatistics
Integrative biology for biomarker discovery
The NEPHRODIA Project
Collaboration LIA, CBS Sfax, Tunisia
Natural history and Diagnosis
Normal albuminuria Microalbuminuria Macroalbuminuria ESRD
Diabetesdiagnostic
diabetes
DN diagnostic
5-10 yrs 5-7 yrs5-10 yrs
Diabetic Nephropathy
Aims of the project• a better understanding of the role of Nephrin
and related proteins in the glomerular filtration process
• the identification of early DN biomarkers by using proteomic approaches
HypothesisProteins that are excreted in the urine precede albumin
MethodsCohort of DT1 patients and DT1 patients at-risk of DNComparative 2D GEL of urine and further identification of biomarker proteins
• ResultsA standardized protocol for urinary proteinspreparation has been set-upA study of the variability of the normal urinaryproteome has been conductedA database of proteins of the normal urinaryproteome has been constructed
DN biomarker discovery
Structural and functional analyses of key molecules in ND
Structure-functionanalyses of proteinnetworks involved in podocyte interactions and filtration
Molecular modeling of Nephrin proteinfamily and associated proteins
Combination of molecular results with kidneysimulations
Nephrodia
Step A : Heterogeneous biological
data recruitment, production, formalization
Step B: Heterogeneous graph
representation and analysis
Step C Identification of key
elements for early onset of disease
Step D : Multi-scale modelling of key elements influence on diseases associated
mechanisms
Step E validation
Biomarkers
Dynamic modelssimulations
Clinicalsamples
HeterogeneousGraphs
Clinicaldata
Proteomics
Genomics
Semi-Automatic Literature mining
Biomarker discovery loop
Combination of bio-experiments, literature mining and modeling approaches
Biomarker discovery loop
ClinicalSamples
Biomarkeridentification
High-throughput Complementary
Antibody pair design
Multiplexed assay
development
Genomics
Proteomics2D vs LC-MS
Knowledge base
Biophysic
010001011101010110010011101
Biological network Modelling, Simulation
Systems biology and discovery
Expert
UnderstandingObservation Engineering
DataInformation
Modelling
Control
Design
Synthetic BiologyRational vs Systematic ways
Biochemistry
protein-gene-RNA interactions
protein-protein interactions
PROTEOME
GENOME
TRANSCRIPTOME
Citrate Cycle
METABOLISM
Bio-chemical reactions
Multi-scale modelling
What do we know from biology
Biological data are partialNot always « true »
Not always understood
Biological networks are complexinter-dependent
dynamicredondent/robust
Scale-free ?
Molecular functions are complexregulated (+/-)
pléiotropicstochasticdynamic
Keyword : « it depends »
Keyword : « who knows ? »
Keyword : « nobody knows...»
Modeling objects or actions ?
The Time problem
The state problem
The Location problem
Modeling objects or actions ?
Bairoch A (1993). The ENZYME data bank. Nucleic Acids Res., 21, 3155-3156
Enzyme Classification E.C.
http://www.chem.qmul.ac.uk/iubmb/enzyme/
Nomenclature effort (~3800 E.C.)Linked to 3D structures
Problems :Close to biochemical point of view, far from cellular Pt of viewReduced to enzymatic processesBio-Object orientedDoes not address multifunction protein
Gene ontology
Multi-scale
Problems :Bio-object orientedDescription of THE function of the molecule.Fixed number of possible descriptions (sensitive to knowledge changes)No easy link to 3D structure
http://www.geneontology.org/
Ambiguité sur la fonction
Récepteur du FGF (facteur de croissance)
Quelle fonction pour cette protéine ?
KinaseDomain
FGF
GO:0003673 : Gene_Ontology (103367)GO:0008150 : biological_process (68451)
GO:0009987 : cellular process (26824)GO:0007154 : cell communication (7527)
GO:0007165 : signal transduction (5890)GO:0007166 : cell surface receptor linked signal transduction (2414)
GO:0007167 : enzyme linked receptor protein signaling pathway (701)GO:0007169 : transmembrane receptor protein tyrosine kinase signaling p
GO:0008543 : FGF receptor signaling pathway (54)GO:0005575 : cellular_component (54946)GO:0003674 : molecular_function (75116)EC 2.7.1.37
BioΨ biological processes description scheme based on elementary actions
A new approach :
Biological processes can be described independently from biologicalObject description.
Multi-scale biological process description could be compatible withA multi-scale component structure description
Does exist a limited set of elementary processes which combinedcould describe the biological function diversity ?
Protein are: multi-domainmulti-action
Kinase activities
FGF binding
Heparin binding
We need multi-scale views on biological functions
Molecular functions are dynamic and depend on their environment (biological context)
Biological Activities
Biological functionalities
Basic Elements of action(BEA)
Biological Roles cellular
molecular
Sub-molecular
Biochemistry whichleads to action
Angles of view on biological processesLevels of abstraction
-BEA refer to the elementary actions at a chemical level involved in biological processes.
-Biological Activities represent the use of a combination of BEA byfunctional domains to exert their activity at a sub-molecular level.
-Biological Functionalities represent the integration of the Biological Activitiesof molecular entities.
-Biological Roles represent the combination of the Biological Functionalitiesof different molecular entities within functional modules in the cell.
BioΨ elementary bricks:
ABond Modifiers
a split/linka acting on C-Cb acting on C-Oc acting on C-Nd acting on C-Se others B
Transfertsa Transferors
a acting on C-Cb acting on C-Oc acting on C-Nd acting on C-Se others
b Oxidoreductorsa acting on C-Cb acting on C-Oc acting on C-Nd acting on C-Se acting on Sf acting on N-Og acting on S-Oh others
CIntramolecular modifications
a Isomerorsa chiralityb cis/transc bond movesd others
DNon covalent interactions
a Binding:a Protein-Proteinb Protein-Nucleic Acidc Protein-Otherd Nucleic Acids-Nucleic Acidse Nucleic Acid-Others
b Transport:a Tunnelb Cargo
BEA : Classification of 97 basic elements of action (for all known processes)
Insulin
C3G
SHCIRS1SHPS1
CrkII
Grb2
SOS
ERK1
GSK3
IR
MEK1 MNK1
mTOR
PDK1
PKB
PKCζ
RSK
LAR
PP1CPP2A
PTENPTP1B
SHIP
SHP2
p85αp110α
PI3K
Glycogenesis
Rap1
Raf1
Rap1
Glucose uptake
Transcriptionregulation
NckIRS1
SOS
Ras
14-3-3
Raf1
cytoplasm
extracellular
Protein bindingNucl Ac binding kinase
phophatase
Biological Activities bricks
picosec
nanosec
micosec
second
minutes
hours
t scale
A
nm
micron
mm
cm
Dist scale
Molecular description Processes descriptionElementary bricks of biological processes
J. Mol. Biol. 2004, J. BioSc. 2007
Scales Inter-dependenciesP(l): Biological ProcessA(l): biological actorl = level of abtractionP(l)
Composed of i P(l-1) performed by j A(l-1)
Conditionalities on the i P(l-1)SequentialityLocalizationbiochemical stateconformational state
Kinetics of P(l-1) performs by A(l-1) in a context of SequentialityLocalizationbiochemical stateconformational state
l>2
l>2
[always/never][before/after/while]boolean
BEA classes comparison
0
10
20
30
40
50
60
70
80
90
classA classBa classBb classC
E coliBacillusPombeHeamophimycobactArabidopS. CerevisC elegFruit flyMouseratBovhuman
bond
mod
ifica
tion
Grou
p tra
nsfe
rt
chem
ical r
eorg
anisa
tion
prot
on/el
ectro
ntra
nsfe
rt
BEA relative distribution among species
0
5
10
15
20
25
30
35
Aa:
CC
Aa:
CO
Aa:
CN
Aa:
CS
Aa:
mis
c
Ba:
CC
Ba:
CO
Ba:
CN
Ba:
CS
Ba:
NO
Ba:
PO
Ba:
SO
Ba:
SS
Ba:
mis
c
Ba:
lab
% u
tiliz
atio
n
0
510
1520
25
3035
Bb:
CC
Bb:
CO
Bb:
CN
Bb:
N
Bb:
S
Bb:
NO
Bb:
SO
Bb:
ion
Bb:
mis
c
Bb:
etra
ns
Ca:
chir
Ca:
ct
Ca:
btra
ns
Ca:
mis
c
BEA codes
% u
tiliz
atio
n
Escherichia coli Bacillus subtilisHaemophilus influenzae Mycobacterium tuberculosisSchizosaccharomyces pombe Saccharomyces cerevisaeCaenorhabditis elegans Arabidopsis thalianaDrosophila melanogaster Mus musculusRattus norvegicus Bos taurusHomo sapiens
Ca
A: Bond Modifications B: Transferts C: Chemichal reorganisation D: No Chemichal Modification
BEA relative distribution among species
BioΨ what for ?
Formalized functionnal description common to all species
Biological function knowledge base.
Component
Process(Conditions)
Localization(Constraints)
Orthologues
Protéines AutresComposantsAcides NucleiquesProcessus Localisation > Espèce
SiliBase Ver. β1.0
Mapping structure-function relationships(support for 3D structure or function predictions, Design of new functions ?)
BioΨ what for ?
limited # folds in PDB
Different Biological FunctionalitiesSimilar biological activities
Orthogonal bundle fold
Functional comparison between organisms
BioΨ allows processes comparison without sequence comparison.
BioΨ overcomes the problem of :Emergent processesDomain shufflingEtc.
BioΨ what for ?
Modeling and simulation of biological processes
Elementary bricks of process at each level can be consideredas primitives for formal language construction.
Genericity of process description is an advantage for nonDeterministic modeling
Easy to use with cellular automaton and multi-agents approaches
On going works : MitochondriaB. Subtilis (EU BaSysBio)Synthetic Biology
BioΨ what for ?
15 Biological Functionalities11 Biological Activities16 Basic Elements of Action
Generic TCA Cycle : a case study
Pathway comparison without sequence analysismulti-agent simulations
BA_thioacyl_thiol_transferase BA_thioacyl_thiol_transferase using Input1 and Input2 to obtain Output1 and Output2(Ba:CS.2 with C-S-R==Input1 andwith C-S-R==C(=O)-S-Rwhile Ba:lab.2 with R-H==Input2 andwith R-H==R-SH andwith R-H==R-CH(R')-SH )and always after ( Ba:lab.2 back with R-H==Output1while Ba:CS.2 back with C-S-R==Output2)
BA_thio_oxidase BA_thio_oxidase using Input1 and Input2 to obtain Output1 and Output2Bb:S.1 with R-SH==Input1 and
with R'-SH==Input1 andwith R-S-S-R'==Output1
and always after Bb:CN.1 back with R-N=C(R')-R''==Input2 and always after Ca:btrans.2 with R-C(=N-R')-R''==Output2
BA_FADH2_NAD_reductase BA_FADH2_NAD_reductase using Input1 and Input2 to obtain Output1 and Output2Ca:btrans.2 back with R-C(NH-R')=R''==Input1and always after Bb:CN.1 with R-N=C(R')-R''==Output1and always after Bb:CC.1 back with R-C(R')=C(R'')-R'''==Input2 and
with R==C(R)-C(=O)NH2and always after Ca:btrans.5 with R-C(=R')-N(R'')-R'''==Output2
BA_thioacyl_keton_transferase BA_thioacyl_keton_transferase using Input1 and Input2 and Input3 to obtain Output1 and Output2(Ba:CS.2 with C-S-R'==Input1 while Ba:lab.1 with H-OH==Input2)and always after (Ba:lab.2 back with R-H==Output2 and
with R-H==R-S-Hwhile (Ba:CO.1 back with C°==Interm1while Aa:CC.2 back with R-C(OH)(R')-R''==Output1 and
with R-CO-R'==Input3 andwith R'-H==Interm1 and with R'-H==R'-O-H ))
BioΨ formal description
Central Carbon metabolism (Gluc/Mal shift)
BioΨ formalization,Modules identification, Context variation and spacio-temporal simulations
Formalization of structural and functional knowledgeFrom molecular details to network
ExperimentalParameters
Multi-scaleconstraints
Multi-scalestructural organization
Model Repository
SBML
BioΨ language
Multi-agents systemDynamic and spacial
simulation
Flux analysesElementary modes
Multi-scale modelling
Model checkingRobustnessDynamic modellingODEEtc.
Complex System Modeling and Engineering for Diagnosis
CompuBioticA synthetic biology approach for new Diagnostic devices : CR Cancer
Complex System Modeling and Engineering for Diagnosis
Synthetic biologyA) the design and construction of new biological parts, devices, and systemsB) the re-design of existing, natural biological systems for useful purposes.
Zauner , 2006.
Design network of biomolecule able to realize elementary tasks which, whencombined perform « programmed » processes.
Our goalDesign and build a synthetic Vesicule « programmed » to perform
in vitro or in vivo diagnostic assays
Colo-rectal cancer diagnosis and follow-up
« Close to the patient » simple assay
Multi-parametric assay
Sophisticated signal integration(qualitative, quantitative, temporal, spacial etc.)
Result return in a simple way (local dying)
- The test can detect abnormalities only in the lower part of the rectum. - Additional procedures are necessary if the test indicates an abnormality.
- Often part of a routine physical examination. -No preparation of the colon is necessary. - The test is usually quick and painless.
Digital Rectal Exam (DRE)
- The test may not detect some small polyps and cancers. - Thorough preparation of the colon is necessary before the test.- False positive results are possible. - The doctor cannot perform a biopsy or remove polyps during the test. - Additional procedures are necessary if the test indicates an abnormality.
- This test usually allows the doctor to view the rectum and the entire colon. - Complications are rare. - No sedation is necessary.
Double Contrast BariumEnema(DCBE)
- The test may not detect all small polyps and cancers, but it is the most sensitive test currently available. - Thorough preparation of the colon is necessary before the test.- Sedation is usually needed. - Although uncommon, complications such as bleeding and/or tears in the lining of the colon can occur.
- This test allows the doctor to view the rectum and the entire colon. - The doctor can perform a biopsy and remove polyps during the test, if necessary.
Colonoscopy
- This test allows the doctor to view only the rectum and the lower part of the colon. Any polyps in the upper part of the colon will be missed.- There is a very small risk of bleeding or tears in the lining of the colon. - Additional procedures, such as colonoscopy, may be necessary if the test indicates an abnormality.
- The test is usually quick, with few complications. - Discomfort is minimal. - In some cases, the doctor may be able to perform a biopsy(the removal of tissue for examination under a microscope by a pathologist) and remove polyps during the test, if necessary. -Less extensive preparation of the colon is necessary with this test than for a colonoscopy.
Sigmoidoscopy
- This test fails to detect most polyps and some cancers.- False positive results are possible. ("False positive" means the test suggests an abnormality when none is present.) - Dietary and other limitations, such as increasing fiber intake and avoiding meat, certain vegetables, vitamin C, iron, and aspirin, are often recommended for several days before the test. - Additional procedures, such as colonoscopy, may be necessary if the test indicates an abnormality.
- No preparation of the colon is necessary. - Samples can be collected at home. - Cost is low compared to other colorectal cancer screening tests. - FOBT does not cause bleeding or tears in the lining of the colon.
Fecal Occult Blood Test (FOBT)
DisadvantagesAdvantagesTest
Table: Advantages and Disadvantages of Colorectal Cancer Screening Tests
Need something new !Blood test CA19-9, CEA etc. poor specificity
Test Hemocult® simple mais, peu sensible, peu spécifique
Coloscopie lourde à mettre en œuvre et peu sensible
Pas de test periphérique spécifique
Scientific contex
Il existe pourtant des biomarqueurs de surface ou sécrétés
Colo-Rectal CancerResponse of the patients to a treatment
First set of biomarkers to predict patient response to Campto®
Responders Non-respondersResponders Non-responders
RespondersNon-responders
RespondersNon-responders
RespondersNon-responders
(Coll. CRLC,Aventis/Pfizer)
International patent CNRS/PfizerJ.Clin.Onco. 2007, Cancer Research 2008, Mol. Cancer 2008
Main strategies in synthetic biology
Bromley et al ACS Chem biol.2008
Ron Weiss, MIT, Harvard,Design of a lentivirus able to target breast cancer intra-cellular biomarkers RNAi networks
Steem cell reprograming by bact. for tissus reconstructionPL Luizi, Roma autocatalytic vesicues constructions, minimal cells.J. Stelling ETHZ, Zurich Electronic-like circuit design with compasable parts (Bact.)V. Dos santos, Helmotz Inst. Germany reprogrammed bact. to target cancer cell
Pseudomonas putida.Luca Cardelli , Microsoft Research, Cambridge, Uk.Cell automaton, semantic of collective behaviourJim Haseloff, Univ Cambridge UK , Plant reprogramming
Alfonzo Jaramillo, E. Polytechnique, reseau RNAi (théorique), proteine design.Antoine Danchin, Inst Pasteur. (Bact)
Design – Simulation - Experimentalvalidation
BioΨFormalization
SimulationStochastic Cellular automaton, Multi-Agent
HsimFunctional bircks
Modelling:Control Th.flux analyses
ODEElementary modes
Design synthetic network
E1
E3
S1
P1
P3
E4
E5
S2
P2
P4
P5
P4P3
Biotechnology
Auto-organisation/Robustness
Validation in vitro
Identification and characterization of molecular compounds
in vitro Proof of concept
Notre Stratégie
Standardized catalog of proteic biological compounds:Processes are formalized (ready for modelling)Biological behavior characterized experimentaly (ready to use in a synthetic system
« compound » propertiesrobustnessstabilityfunctional diversity relative to context modifications
Design
System design using our compound catalogSBGN and Celldesigner
BioΨ modellingFormal description
Simulation : Stochastic Cell automatonand multi-agent
Simulation
Construire le réseau biologique synthétique
Simuler son comportement à l’échelle.
50 nm
Experimental validation
Stable Vesicles construction (liposomes) ~100nm
Introduction of chosen functional compounds (GOD-POD)
Opérational assays of full synthetic system (in vitro)
SP
S
P
S
Experiments• Enzymatic system : GOD-POD (kit Sigma GAGO20)
• Glucose oxidase – peroxidase• Reactions :D-glucose + O2 + H2O D-gluconic acid +H2O2H2O2 + reduced o-dianisidine (colorless) -> oxidized o-dianisidine (brown)
D-glucoseGlucose oxidase
O2H2O
D-gluconic acidH2O2
reduced o-dianisidine(colorless)
oxidized o-dianisidine(brown)
Peroxidase
Simulation
Vesicles Construction
2 Liposomes formulation: Lipid film hydration:
1 Injection Lipides/ethanol in H20 Solution
Sabine PérèsStephanie RialleFranck MolinaLiza FelicoriAlain ThierryDavid JeanChams KifagiCecile FleuryNicolas SalvetatEve DupatViolaine MoreauKarine Kaminski
A Jaramillo, E. polytechnique, Palaiseau, Fr P Noirot, INRA Jouy-en-Josas, FrJ Stelling, ETHZ, Zurich, CHJ Banga, CSIC, Vigo, SpainB Schickowski, Inst. Pasteur, Paris, FrP Amar , LRI orsay, FrF. Képes, Genopole Evry, FrE. Klipp, MaxPlanck Inst.,Berlin DF. Fages, INRIA
Collaborators
Montpellier, France
Groupe National de Biologie Synthetique, Genopole Evry®A. Jaramillo, F. Molina, F. Fages
EU, BaSysBio
18 academic staff (8 permanents)9 CNRS, 2 INSERM, 2 PostDoc, 5 phD
23 Bio-RadBiorad 1 RdS, 4RdP, 9CdP, 4AcP, 4T, CDD 1CdP
5 undergraduates70% wet lab experimentalists 30% theoreticians