1. From Molecule to Man: the Virtual
Physiological Human, Computational
Biology & the Future of Biomedicine
Peter V. Coveney
Centre for Computational Science
University College London
2. VPH and Computational Biology Issues
Presentation Outline
• Context
• VPH Initiative
– VPH Projects
– Role of VPH NoE
• Centre for Computational Science Research
• UCL/Yale Partnership
• “Digital Me” – a vision of future healthcare provision
3. VPH/Physiome History
Human Genome Project
Systems Biology
Grid Computing
Finite Elements
Microcomputers/home FP7 call 2
computers Objective
ICT-2007.5.3:
Molecular Biology Virtual
Physiological
Human
1993 1997 2005 2006 2007 2008 2009
4. VPH and Computational Biology Issues
Integrative and systems biology
• Data produced by sequencing 10TB per day per machine
• Integrating over all relevant length and timescales
• Requires integration of suitable compute and data facilities,
connected by high performance networks.
• VPH seen as paradigm case for putative BBSRC Digital
Organisms programme.
• Exploiting ‘big data’ – seizing new opportunities to generate
knowledge and impact as we enter an era of data intensive
science.
• Tools, resources and facilities – providing the tools,
technologies and infrastructures that are essential for 21st
century bioscience.
• Increase the uptake of systems-based approaches in the private
sector.
5. VPH and Computational Biology Issues
Funding: Budget for research supported by the European
Commission ICT - Health unit in FP7
Overall € 340 M over 4 years (2007-2010) Challenge 5
• Predictive Medicine–Virtual Physiological Human (VPH
Initiative)
– Modelling/simulation of human physiology and disease € 72
M in 2007 - € 5 M in 2009 - € 68 M in 2010
• Personalisation of Healthcare
– Personal Health Systems (PHS), € 72 M in 2007 - € 63 M in
2009
• Patient safety (PS) - avoiding medical errors
– € 30 M in 2007 - € 30 M in 2009
6. VPH and Computational Biology Issues
The VPH Initiative is currently in its 3rd Call. It focuses on:
• Keep the overall VPH vision:
– Early diagnostics & Predictive medicine
– Personalised (Patient-specific) healthcare solution
• By means of:
– Modelling & simulation of human physiology and disease related
processes
– Emphasis on tools/infrastructure for bio-med researchers+clinicians
• VPH in 2010
– Call 6 opens November 2009 , closing 13 April 2010
– Budget € 63M
7. VPH and Computational Biology Issues
Overall planning for the elaboration of the
EU Workprogramme 2011-12:
• Orientation paper developed by Christmas 2009
• First draft of the Workprogramme by February 2010
• Final draft of the Workprogramme by May 2010
8. VPH and Computational Biology Issues
• Elaboration of WP2011-12:
– Drafted by the Commission
– Based on community inputs (research/clinical/industry)
• The NoE Vision Document is an essential input for the VPH
– It presents the view of the Community at large on gaps,
challenges and future direction and informs EC policy
• There will be annual updates with open consultation in the
community through the NoE website.
• Don’t miss this opportunity to contribute!
9. VPH and Computational Biology Issues
The VPH Initiative (VPH – I) & VPH Network of Excellence
• Collaborative projects within the call meet objectives
associated with specific challenges
• VPH NoE connects all of these projects, and must focus on
addressing issues of common concern that affect VPH-I
projects collectively
– research infrastructure
– training
– dissemination
10. Virtual Physiological Human
• Funded under EU FP 7 (call 2)
• 15 projects: 1 NoE, 3 IPs, 9 STREPs, 2 CAs, and new
projects in negotiation phase
“a methodological and technological framework that, once
established, will enable collaborative investigation of the
human body as a single complex system ... It is a way to
share observations, to derive predictive hypotheses from
them, and to integrate them into a constantly improving
understanding of human physiology/pathology, by regarding it
as a single system.”
VPH Call 6 currently open – deadline13th April 2010
11. VPH- I FP7 current projects
Industry Parallel VPH projects
Grid access CA
CV/ Atheroschlerosis IP Liver surgery STREP
Breast cancer/ diagnosis
Heart/ LVD surgery STREP
STREP
Osteoporosis
Oral cancer/ BM IP
D&T STREP
Networking
Heart /CV disease NoE Cancer STREP
STREP
Vascular/ AVF & Liver cancer/RFA
haemodialysis STREP therapy STREP
Heart /CV disease Alzheimer's/ BM &
STREP diagnosis STREP
Security and Privacy in
VPH CA
Other
Clinics
13. VPH and Computational Biology Issues
Specifically, the VPH NoE will:
• Identify user needs, define standards, ontologies and
applications, and develop VPH ToolKit
• Develop VPH training activities and materials: Joint advanced
degree programme, interdisciplinary study groups, focused
journal issues, textbook
• Provide research/news dissemination services and international
EU/international networking
Project Coordinators: Vanessa Díaz-Zuccarini / Miriam Mendes (UCL)
Scientific Coordinators: Peter Coveney (UCL), Peter Kohl (Oxford)
http://www.vph-noe.eu
14. VPH and Computational Biology Issues
• 14 Core Partners
– 4 UK (UCL, UOXF, UNOTT, USFD)
– 3 France (CNRS, INRIA, ERCIM)
– 2 Spain (UPF, IMIM)
– 1 Germany (EMBL [EBI])
– 1 Sweden (KI)
– 1 Belgium (ULB)
– 1 New Zealand (UOA)
– 1 Italy (IOR)
• Associate / General Members
– 38 General (academic) and Associate (Industrial) Members
– … and growing
15. VPH and Computational Biology Issues
Future results and impacts
• Create a more cohesive VPH research community,
both within and beyond the EU.
• Enhance recognition at national level of importance of
modelling and simulation in biomedicine
• Increase Industrial and Clinical awareness of VPH
modelling
• Increase emphasis on interdisciplinary training in
biological and biomedical/engineering physics curricula
16. VPH and Computational Biology Issues
Project Structure
WP4: WP5:
INTEGRATION SPREADING
AND TRAINING ACTIVIES EXCELLENCE
WP2 - Exemplar Projects
Vertical integration
WP1 - Management
WP3 - VPH ToolKit
17. VPH and Computational Biology Issues
Project Structure
WP2 – Exemplar Projects
WP5 - Networking/Communication and
Spreading Excellence within the VPH
NoE/VPH-I
WP1 - Management
WP4 – Integration &
WP3 – VPH Toolkit
Training activities
18. WP1
VPH NoE General Assembly - Key Legislative Body
(Includes all membership types)
Annual Meetings (coinciding with Project Meetings)
Delegates Executive Powers to Steering Committee
Steering Committee: Clinical
Advisory Board
Key Executive Body
Includes Consortium Agreement Signatories Industry
Quarterly meetings to lead implementation of NoE WPs Advisory Board
Takes advice from Advisory Boards
May set up Task Forces on Policy Issues (such as on IP, Ethics, Gender)
Scientific
Delegates day-to-day project management to WP1 - Management Advisory Board
Project Office
UCL European Research & Development Office EC Project Officer
Contract management and project implementation
WP2 WP3 WP4 WP5
Other VPH Other VPH Other VPH Other VPH Other VPH
Projects Projects Projects Projects Projects
30. Biomedical application have
“non-standard” requirements
• Ability to co-reserve resources (HARC)
• Launch emergency simulations (SPRUCE)
• Uniform interface for federated access
• Access to back end nodes: steering, visualisation
• Lightpath network connections
• Cross site simulations (MPIg)
• Support for software (ReG steering toolkit etc)
31. These requirements impact resource
provider policies
• TeraGrid, NGS & HPCx starting to support advanced
reservation with HARC
• DEISA evaluating HARC deployment on their systems
• Some TeraGrid sites support emergency jobs with
SPRUCE
• Lightpath connections in place between Manchester -
Oxford NGS nodes and Manchester - UCL
• MPIg and RealityGrid steering deployed on NGS and
TeraGrid resources
32. CCS -HIV drug design/drug treatment
HIV-1 Protease is a common target for HIV drug therapy
• Enzyme of HIV responsible for
Monomer B Monomer A
protein maturation 101 - 199 1 - 99
• Target for Anti-retroviral Inhibitors Flaps
• Example of Structure Assisted Drug Glycine - 48, 148
Design
• 9 FDA inhibitors of HIV-1 protease Saquinavir
So what’s the problem?
• Emergence of drug resistant
mutations in protease
• Render drug ineffective P2 Subsite Catalytic Aspartic
• Drug resistant mutants have Acids - 25, 125
emerged for all FDA inhibitors Leucine - 90, 190 C-terminal N-terminal
33. CCS - Determination of protein-drug binding affinities
Binding of saquinavir to wildtype and
Simulation and calculation workflow resistant HIV-1 proteases L90M and
G48V/L90M
Applications used include: NAMD,
CHARMM, AMBER… Thermodynamic decomposition
• explains the distortions in enthalpy/entropy
Rapid and accurate prediction of binding free energies for balance caused by the L90M and G48V
saquinavir-bound HIV-1 proteases. mutations
Stoica I, Sadiq SK, Coveney PV.
J Am Chem Soc. 2008 Feb 27;130(8):2639-48. Epub • absolute drug binding energies are in excellent
2008 Jan 29. agreement (1 – 1.5kcal/mol) with experimental
Doable in days values
34. Automation of binding affinity calculation
• Eventual aim is to provide tools that allow simulations to be used in a clinical
context
• Require large number of simulations to be constructed and run automatically
– To investigate generalisation
– Automation is critical for clinical use
• Turn-around time scale of around a week is required
• Trade off between accuracy and simulation turn around time
Binding Affinity Calculator (BAC)
A distributed automated high throughput binding affinity calculator for HIV-1
proteases with relevant drugs
“Automated molecular simulation based binding affinity calculator for ligand-bound HIV-1 proteases”. Sadiq SK, Wright D, Watson
SJ, Zasada SJ, Stoica I, Coveney PV. J Chem Inf Model. 2008 48(9):1909-19.
35. Grid enabled neurosurgical imaging using simulation
The GENIUS project aims to model large scale patient specific cerebral blood flow in
clinically relevant time frames
Objectives:
• To study cerebral blood flow using patient-specific image-based models.
• To provide insights into the cerebral blood flow & anomalies.
• To develop tools and policies by means of which users can better exploit
the ability to reserve and co-reserve HPC resources.
• To develop interfaces which permit users to easily deploy and monitor
simulations across multiple computational resources.
• To visualize and steer the results of distributed simulations in real time
Yields patient-specific information which helps plan embolisation of arterio-venous
malformations, aneurysms, etc.
37. Modelling vascular blood flow - HemeLB
Efficient fluid solver for modelling brain bloodflow
• Uses the lattice-Boltzmann method
• Efficient algorithm for sparse geometries
• Machine-topology aware graph growing partitioning technique,
to help minimise the issue
of cross-site latencies
• Optimized inter- and intra-machine
communications
• Full checkpoint capabilities.
M. D. Mazzeo and P. V. Coveney, "HemeLB: A high performance parallel lattice-Boltzmann code for large scale fluid
flow in complex geometries", Computer Physics Communications, 178, (12), 894-914, (2008).
38. CCS - Haemodynamic simulation and visualisation
• First step is the conversion of MRA or 3DRA data (DICOM format) to a 3D model,
vasculature is of high contrast, 200 - 400 µm resolution, 5003 - 10003 voxels
• 3DRA - 3-dimensional rotational angiography, vasculature is obtained using digital
subtraction imaging with a high-contrast x-ray absorbing fluid.
• Each voxel is a solid (vascular wall), fluid, fluid next to a wall, a fluid inlet or a fluid outlet
• Arterial simulations typically have 1-3 inlets and ~50 outlets,
• We apply oscillating pressures at inlet and oscillating or constant one at the outlets, for
example.
• Real-time in-situ visualisation of the data using streamlines, iso-surfacing or volume
rendering (stress, velocity or pressure).
• New insights into physiology of human brain
Reconstruction and boundary condition
Velocity field obtained with
set-up; fluid sites, inlet and outlet sites in
our ray tracer
red, black and green respectively;
HemeLB
38
39. Translational Impact
Book computing resources in advance or have a
system by which simulations can be run urgently.
Move imaging data around quickly over
high-bandwidth low-latency dedicated links.
Interactive simulations and real-time
visualization for immediate feedback.
15-20 minute
turnaround
39
40. Yale – UCL Collaboration
UCL-Yale Computational Biomedicine collaboration
• Aims: Pool the computational, information management and
analysis resources of two of the world’s greatest universities
and their associated hospitals:
– Yale University Hospital
– UCL Partners (Great Ormond Street, Moorfields Eye Hospital, Royal Free
Hospital and UCL Hospitals Trust).
• break-through science and speed up its translation to better
medicine.
• Far-reaching transformative
consequences for UK and US science,
medicine and industry.
41. Yale – UCL Collaboration
• Overview of Yale- UCL Collaboration
• World Ranking Yale 3, UCL 4.
• Framework Agreement signed 8 Oct 2009 by Malcolm
Grant & the President of Yale Richard Levin & leaders
of associated hospitals
• Joint resources committed to research & education
with aims:
– Enhance scope & quality of research both basic & clinical
– Attract the best students to a joint PhD programme
– Drive translation within the Collaboration
– Lead improvement in patient care
43. Yale – UCL Collaboration
Achievements (first 12 months)
Bottom up approach
Basic research: 10 joint projects started
Clinical research: 2 UCL research clinics started in Yale
Hospital, 1 Yale clinic in UCLP. (1st NIH grant $4.7m).
Bioinformatics: “The Coveney Plan”
Drug Discovery: 3 projects (1 patent)
First into Man: First project completed
Hospitals: Exchange of managers
44. Yale – UCL Collaboration
Future
• Enthusiasm across Biomedicine & Life Sciences
• Establish translation with Collaborative
• Joint fund raising
• Split PhD starts September 2011
• Joint Hospital Management Institute
• Governance running
45. Yale – UCL Collaboration
UK, EU and Global Benefits of the Collaboration
• Major contribution to UK Digital Economy - positioning the UK at the centre of globally
integrated healthcare data management
• Strengthening UK expertise in information management
– Expanding market, skills and employment opportunities in information management
• New paradigm will transform healthcare in the UK
– Improving health and well being of UK citizens through better patient data management
– Bespoke treatments and specific drugs for the patient using digital information and clinical expertise
– Will help contain health costs and improve patient wellbeing
• Develop and invest in these future technologies now
– No unified management of & funding mechanisms for data, network and computing
– Current funding is fragmented - networks (JISC), computing (EPSRC, STFC) and data (BBSRC)
– Investment case in point: EBI was placed in the UK owing to superior networking technology which was
invested in ahead of time
– UK HPC likely to be embedded within EU via DEISA/PRACE
– UCL is in an excellent position with strong links to EBI, NIH, Sanger, Wellcome, MRC & BBSRC
(including planned Digital Organism programme)
46. Yale – UCL Collaboration
UCL-Yale integrated technological platform
• Current and future biomedical research that the UCL-Yale
Collaboration will tackle all share the common themes of being
compute- and data-intensive
– The effective integration of distributed data, computing and networks is central
to success
• Data: Standards, transparency, security and ethics
• Computing: Access to high-performance computational resources in
UK, EU and USA
• Networks: Establishment of persistent high-bandwidth, high QoS
connections between:
– Data producers (e.g. Sanger, EBI, …)
– Researchers (UCL, Yale)
– Clinicians (UCLP, Yale Hospital)
– Compute resources (DEISA, TeraGrid e.g.), initially at 10Gbs, rising to 100Gb/s
and to 1Tb/s by 2020
48. Yale – UCL Collaboration
UCL-Yale Collaboration in Biomedicine
• The Collaboration has already established joint projects which
are totally integrated
• Cardiovascular science is the first discipline to have been
integrated over the last year
• First joint grant was awarded from the NIH in September 2009
(genetics of congenital heart disease)
• Women’s health, cancer, neuroscience and surgery will be
integrated during 2010
49. Yale – UCL Collaboration
UCL-Yale Collaboration in Biomedicine:
Examples
• Cardiovascular clinical genetics - whole genome high throughput
analysis at Yale is being applied to UCL patient databases.
• Cardiovascular imaging – Combined power of multiple imaging
modalities available within the Collaboration, for patients, volunteers
and large and small animals, is world-leading.
• Cardiovascular research clinics
– UCL staff have established clinics for congenital heart disease and familial sudden
death at Yale
– Yale staff have established a program for treatment for chronic total occlusion of
coronary arteries within UCLP.
50. Digital Me
What role will integration play in medicine?
• Up to 1990: anecdote-based medicine (based on personal
experience of the physician)
• 1990-2010: evidence-based medicine (based on consensus
treatment protocols derived from population studies)
• 2020: truly personalised medicine, based on models derived
from individual patient profiles (genotypes and phenotypes)
• Making decisions from these rich data will require huge amounts
of computation, as well as integration of network, computing,
and data resources – globally.
51. Digital Me
A vision of future healthcare
• Personalised, Predictive, Integrative...
• An integrated anatomical, physiological, biochemical,
genotypical model of me
• Integrates my health “chronicle” with models derived from
populations with similar phenotypes
• Lets me understand and take control of my own life and health –
“$2000 genome” in 3-5 years?
• Used for:
– Patient education
• “what is happening to me?”
• “what are you going to do to me?”
– Management of chronic diseases
– Clinical decision support
52. Digital Me
Population
Information
Personal
Information Epidemi- Therapy
Genotypes
ology outcomes
Lifestyle
Visualisations
Devices
Digital Me
Genotype
Sensors
Predictive
Models
Clinical
Many of the individual components of this figure already
exist and work, however they need to 52
be integrated
53. Digital Me
Between now and 2020
What developments will make Digital Me possible?
• Desk-side supercomputing with 1,000s or even 1,000,000s of cores
willprovide the necessary power to run patient specific simulations at
low cost
• Distributed, ‘grid’, ‘cloud’ or ‘utility’ computing could make the
computational power required for large scale simulations available
transparently, and integrated in to the clinical workflow
• Computational models could be routinely constructed and stored with
a patient’s EHR (electronic health records) to use when needed
– A radical evolution of healthcare
– Encourages individual’s responsibility for own health management of chronic
diseases
– Reduction of need to access central resources
