SIAM Conference on the Life Sciences San Diego, 7-10 Aug 2012 Computational Physiology & the VPH/Physiome Project Peter Hunter Auckland University, NZ, and Oxford University, UK
SIAM Conference on the Life Sciences San Diego, 7-10 Aug 2012
Computational Physiology & the VPH/Physiome Project
Peter Hunter
Auckland University, NZ, and Oxford University, UK
Tissue
Osteon Nephron Acinus Liver lobule Lymph node Cardiac sheets
Organ
Heart Lungs Diaphragm Colon Eye Knee Liver
Environment
Organ system
Organism
Cell
Protein Gene Atom
Network
x 1million 20 generations
The challenge: organs to proteins
• Biophysically based models at every level – as much as possible (there’s always a black box!)
• Adoption of model and data standards – SBML, CellML, FieldML for models
• Automated assembly of multi-scale models – molecule to organ(ism)
• Automated model reduction – otherwise too expensive
• New instrumentation – new instruments → new expts → new knowledge
A multi-scale bioengineering approach needs:
History of the VPH/Physiome Project ..
1999 Systems Biology Markup Language
2006 STEP: Strategy for European Physiome 2008 VPH Network of Excellence
2011 VPH Institute
2010 Drug Disease Model Resources (DDMoRe)
1997 IUPS Physiome Committee
1998 CellML, FieldML
2003 IMAG (NIH, NSF, FDA, NASA, DOE, DOD, ..)
Cardiovascular system Respiratory system Musculo-skeletal system Digestive system Skin (integument) Urinary system Lymphoid system Female reproductive system Special sense organs Central nervous system Endocrine system Male reproductive system
Organ system Physiome Projects
Verification, Benchmarks
Experimental measurements
Model standards: SBML, CellML, FieldML
Model repositories: Biomodels, PMR2
Data standards: DICOM, BioSignalML, ..
Minimum information standards: MIAME, MICEE, ..
Data repositories: PhysioNet, CAPdb, CVRG, ..
Standards for models, data & software
Validation, Limitations
(Journal review)
APIs, webservices
Software: OpenCOR, OpenCMISS, Continuity, Chaste, ..
Simulation standards: SED-ML Functional curation
APIs, webservices
Curation Annotation
Reference description
Metadata: Ontologies
Need modules!
Metadata: Ontologies GO, FMA, ..
CellML – standards, databases and tools
(www.cellml.org)
Cuellar AA, Lloyd CM, Nielsen PF, Halstead MDB, Bullivant DP, Nickerson DP, Hunter PJ. An overview of CellML 1.1, a biological model description language.SIMULATION: Transactions of the Society for Modeling and Simulation, 79(12):740-747, 2003
Cell cycle (25 models)
Calcium dynamics (63 models) Cell migration (2 models)
Circadian rhythms (9 models) Endocrine system (29)
PKPD models (7 models)
Myofilament mechanics (15) Metabolism (35 models)
Electrophysiology (117 models) Excitation-contaction (15 models)
Gene regulation DNA repair (3) Synthetic biology (5 models)
Material constitutive
laws
Metabolism
1. Glycolysis 2. Gluconeogenesis 3. Pentose phosphate pathway 4. Electron transport chain 5. Tricarboxylic acid (TCA) cycle 6. Lipid metabolism 7. Oxidation of odd-chain fatty acids 8. Fatty acid synthesis 9. Cholesterol biosynthesis 10.Phospholipid synthesis 11.Glycolipid synthesis 12.Triacylglycerol synthesis 13.Nitrogen metabolism 14.Pyrimidine nucleotide synthesis 15.Purine nucleotide synthesis 16.Purine nucleotide degradation 17.Conversion of IMP into AMP & GMP 18.Pyrimidine nucleotide degradation 19.Pyrimidine nucleotide degradation 20.Amino acid catabolism 21.Non-essential amino acid synthesis
1. cAMP signalling 2. Calcium signalling - via cADP-ribose signalling NAADP signalling Voltage operated channels (VOCs) Receptor operated channels (ROCs) IP3-Ca2+ signalling (via PLC-PIP2) DAG-PKC signalling (via PLC-PIP2) PI4-5P2 signalling Inositol polyphosphate signalling PI3-Kinase signalling 3. NO-cGMP signalling 4. Redox signalling 5. MAP-Kinase signalling 6. NF-kB signalling 7. Phospholipase D (PLD) signalling 8. Sphingomyelin signalling 9. JAK-STAT signalling 10. Smad signalling 11. Wnt signalling 12. Hedgehog signalling 13. Notch signalling 14. ER stress signalling 15. AMP signalling
Signaling
Data: Initial conditions
Boundary conditions
Population Atlas
(Cardiac, MSK, ..)
Codes: Simulation
Visualisation
Specification of models & solution
techniques
Models from
repository
SED-ML
BioSignalML, DICOM
Comparison with reference
data
Workflow for reference description
Benchmarks
Quality control?
MICEE etc
CellML, SBML & FieldML
PhysioNet
Protein degradation
Cell cycle
Protein trafficking
Calcium transport
Intracell. signalling
Gene transcription
Gene regulation
Protein regulation
Protein synthesis
Cellular metabolism
Electro-physiology
Cell process Molecular Biology
Genes miRNA mRNA
… Proteins
Sequence Structure
PTMs Binding motifs
… Lipids
… Carbo-
hydrates …
CellML & SBML libraries
Model components annotated with
ontologies
Gene networks
Metabolic modules
Ion channels
Signalling modules
Cell receptors
…
Physiome FieldML library
Stru
ctu
re &
ph
ysic
s
Infrastructure for linking Molecular Biology to Physiome
Cell function
Motility
Meiosis
Mitosis
Apoptosis
Contraction
Signalling
Transport
Growth
Adhesion
ECM protein synthesis
Sensing
Cell type Ti
ssu
e m
oti
fs
Illustrate ideas with 4 organs:
1. Circulatory system: Heart 2. Respiratory system: Lungs 3. Musculo-skeletal system 4. Digestive system: Stomach
cell-cell connections
proteins genomic
sequence
amino acid
sequence
torso
Heart physiome: Multi-physics and multi-scale
3D cell
tissue
heart
cellular processes
nm
m
=109nm
Hunter PJ, Pullan AJ, Smaill, BH. Modeling total heart function. Annual Review of Biomedical Engineering, 5:147-177, 2003 LeGrice IJ, Hunter PJ, Smaill BH. Am.J.Physiol. 272:H2466-H2476, 1997
Myocardial activation Ventricular wall mechanics Ventricular blood flow Heart valve mechanics Coronary blood flow Neural control
Torso model
Composite lumped parameter
cell model
Hodgkin-Huxley type ion channel model
Markov ion channel model
3D protein model
Coarse grained MD model
Quantum mechanics model
Molecular dynamics model
Continuum tissue model
Organ model
Discrete tissue structure model
Calcium transport models Myofilament mechanics Signal pathway models Metabolic pathway models Gene regulation models
3D cell model
MRI (100m)
MicroCT (5m)
Confocal light microscopy
(0.5m)
Electron tomography
(5nm)
X-ray diffraction
(5A)
Imaging Modelling
Fluid flow
Reaction-diffusion
Electro-magnetic
Finite elasticity
Partial differential equations (PDEs)
Bayesian network description
Molecular dynamics/coarse graining
Poisson-Boltzmann …
Differential algebraic equations
10-9
10-6
10-3
1m
Tissue
Organ system
Organ
Organism
Cell
Protein Gene Atom
Network
Scale Multi-scale
Cardiac structure & kinematics from histology & MRI
Coronal
slice
Short axis
slice
Young AA. Model Tags: Direct 3D tracking of heart wall motion from tagged magnetic resonance images. Medical Image Analysis 3:361-372, 1999
LeGrice IJ, Young AA, Hunter PJ, Smaill BH. J. Mol Cell Cardiol 32(3) A1, 2000
Tissue level structure
Radiological data
Structural data
Molecular data
Model provides framework for aligning data
Mathematical model
Kim, Cannell & Hunter. Changes in calcium current among different transmural regions contributes to action potential heterogeneity in rat heart. PBMB 103(1):28-34, 2010
-60 -40 -20 0 20 40 60
-20
-16
-12
-8
-4
0
4
mV
I Ca d
en
sity (
pA
/pF
)
-20pA/pF
-60mV 40mV
LV epi
-60 -40 -20 0 20 40 60
-20
-16
-12
-8
-4
0
4
mV
I Ca d
en
sity (
pA
/pF
)
RV
-60 -40 -20 0 20 40 60
-20
-16
-12
-8
-4
0
4
mV
I Ca d
en
sity (
pA
/pF
)
Septum
-60 -40 -20 0 20 40 60
-20
-16
-12
-8
-4
0
4
mV
I Ca d
en
sity
(pA
/pF
)
LV endo
Physiological data
Element fields: undeformed coords X,Y,Z (or l,m,q) deformed coords x,y,z fibre & sheet orientations electrical potential concentration fields for Ca2+, O2 , etc
Element with Hermite basis
Large deformation elasticity theory Finite element method
x1
x2
x3 u , u , u , etc x1 x2
material coordinates
Equations: Integral formulation - Galerkin method - Gaussian quad
Costa KD, Hunter PJ, Wayne JS, Waldman LK, Guccione JM & McCulloch AD. ASME J. Biomech. Eng. 118:464-472, 1996 Nash MP and Hunter PJ. J. Elasticity. 61(1-3):113-141, 2001
Governing equations for solid mechanics
Kinematics: finite strain theory, incompressible tissue Equilibrium eqtns: conservation of mass & momentum Constitutive eqtns: based on fibrous-sheet structure nonlinearly elastic, viscoelastic, porous etc Boundary conditions: displacement or force & contact mechanics Other factors: residual stress, growth & remodelling
Costa KD, Hunter PJ, Wayne JS, Waldman LK, Guccione JM & McCulloch AD. ASME J. Biomech. Eng. 118:464-472, 1996 Nash and Hunter. J. Elasticity. 61(1-3):113-141, 2001
Galerkin finite element method
Tissue level function: passive properties
Hunter PJ, Smaill BH, Nielsen PMF. Biophysical J, 49(2):90a, 1986 Malcolm DTK, Nielsen PMF, Hunter PJ, Charette G. BMMB, 1(3):197-210, 2002 Schmid, H., Nash, M.P., Young, A.A., Röhrle, O., Hunter, P.J. J Biomech Eng, 129(2):279-283, 2007
Axial
tension
Axial strain
sheet axis
fibre axis
sheet normal
fibre
sheet
normal
Epi
Transmural confocal image of rat myocardium
Endo
4 mm
Thermopile arrays
Trabecula
RV inner wall
1 mm
Hunter PJ, McCulloch AD & ter Keurs HEDJ. Prog Biophys Molec Biol 69:289-331, 1998 Niederer, S.A., Hunter, P.J., Smith, N.P. Biophysical Journal, 90(5):1697–1722, 2006
- electrophysiology - myofilament mechanics - metabolism - signalling
Model:
Tissue level function: active properties
radial circumferential
base
apex
Putting it all together
Nickerson, D.P., Smith, N.P., Hunter, P.J. New developments in a strongly coupled cardiac electromechanical model. Europace, 7(2):S118-S127, 2005. Smith NP, Hunter PJ, Paterson DJ. The cardiac physiome: at the heart of coupling models to measurement. Experimental Physiology, 94(5):469-471, 2009. Keldermann RH, Nash MP, Panfilov AV. Modelling cardiac mechano-electrical feedback using reaction-diffusion-mechanics systems. Physica D 238(11-12):1000-1007, 2009.
Fluid flow
Reaction-diffusion
Electro-magnetic
Finite elasticity
Continuum equations
Microstructural model
Constitutive law
Multiscale tissue models
Fluid flow
Reaction-diffusion
Electro-magnetic
Finite elasticity
Continuum equations
Microstructural model
Constitutive law
Mixture theory (collagen, etc)
Growth laws
Cellular signalling
Schmid H, Watton PN, Maurer MM, Wimmer J, Winkler P, Wang YK, Rohrle O, Itskov M. Biomechanics and Modeling in Mechanobiology, 9:295-315, 2010
Modelling growth
Musculo-skeletal system
Load generic models into the anatomical component under study:
Web-accessible database of generic models (+ tissue structure):
Generic models of the joints
Shim VB, Hunter PJ, Pivonka P, Fernandez JW. A multiscale framework based on the physiome markup languages for exploring the initiation of osteoarthritis at the bone-cartilage interface. IEEE Trans Biomed Eng. 58(12):3532-6, 2011
Automatic model generation Need fully automated process for segmentation and mesh fitting
1
1. Load DICOMs of a CT scan
2
2. Identify the femurs in the scan
3
3. Segment the femur surface
4
4. Fit a high-quality mesh to the femur
E.g. for the femur in a CT scan:
Model generation results
• Fully automated (batch process @ 6 min/femur) • >85% success rate on 230 normal CT scans • Continue to improve as apply to database of 20k scans
A typical successful case A typical failure case
Analysis
• Analysis of CT image profiles sampled normal to the surface model
• Spatially varying thickness over the surface is incorporated into the model, enabling correspondent comparisons between individuals or groups
• Thickness information aids the construction of accurate 3-D models of internal structure
• Extract dominant modes for population subgroups
• Automatic estimation of cortical bone thickness
• Sub-pixel accuracy down to ~0.5mm
A multi-scale model of the lung
Respiratory system
Tawhai MH, Clark AR, Donovan GM, Burrowes KS. Computational modeling of airway & pulmonary vascular structure & function: development of a `Lung Physiome'. Critical Reviews in BME, 2011.
Digestive system: stomach
Faville et al. BiophysJ. 96, 4834-4852, 2009. Biophysically based mathematical modeling of interstitial cells of Cajal slow wave activity generated from a discrete unitary potential basis.
We currently have:
• Model & data encoding standards • Public repositories of curated, annotated
& validated models • Reference descriptions of models • Multi-scale, multi-physics algorithms • Framework for stochastic modeling • Carefully managed, open source codes • Benchmark problems • Population atlases
Summary
But we need:
• Better model reduction algorithms • Better ways of handling multi-scale
(as opposed to multiple-scale) • 3D cell models • A comprehensive modelling
framework for metabolic pathways • A comprehensive modelling
framework for signalling pathways • Better links to bioinformatic databases • Better links to medical informatics
Acknowledgements: The CellML/FieldML team
Poul Nielsen
Andrew Miller
Randall Britten
Richard Christie
Alan Garny
David Nickerson
Tommy Yu
Jonna Terkildsen
Mike Cooling
Catherine LLoyd
Musculo-skeletal team Thor Besier Justin Fernandez Alice Hung Duane Malcolm Kumar Mithraratne Katja Oberhofer Vickie Shim Tim Wu Yu Zhang
Cardiac team Bruce Smaill Ian LeGrice Denis Loiselle Martyn Nash Greg Sands Andrew Taberner Alistair Young Jichao Zhao Andrew McCulloch Nic Smith
Lung team Merryn Tawhai Kelly Burrowes Alys Clark
GI team Andrew Pullan (1962-2012) Leo Cheng Peng Du Greg O’Grady
NZ Health Research Council NZ Ministry of Science & Innovation NZ Maurice Wilkins Centre CoRE UK Wellcome Trust (Heart Physiome) FP7 (euHEART & NoE) NIH (Cardiac Atlas Project - CAP)
Acknowledgements: Funding
www.vph-noe.eu