Modelling and Analysis for Biological Systems

1
Modelling and Analysis for Biological Systems

• Marta Kwiatkowska
• Director, Midlands e-Science Centre on Modelling
  and Analysis of Large Complex Systems
• www.cs.bham.ac.uk/mzk
• Converging Sciences Conference
• Trento 16-17th December 2004

2
Complex systems

• Natural and manmade
• Organisms, cells, molecules, particles,
• Global computer, Internet, ., motes, ,
  DNA-bots
• Science drivers
• Curiosity, knowledge, discovery, advancement,
• Prediction, control, well-being,

3
Characteristics of complex systems

• Networks of subsystems
• Bodies, agents, molecules, particles,
• Interaction
• Governed by rules
• Causes transformations
• Evolution
• Continuous and discrete dynamics
• Mobility
• Motion in space and time, reconfigurability
• Stochastic behaviour
• Unpredictability, noise,

Not unlike the global computer
4
Challenges of biological systems

• Scale
• Sheer amount of information
• Number of components
• Multiple levels
• Complexity
• Interaction patterns
• Non-linearity
• Hybrid dynamics, continuous and discrete
• Evolution
• Uncertainty
• Inaccuracy of measurement
• Poor understanding of cause and effect
• Stochastic behaviour

5
Computational foundation

• We argue for
• Rigorous foundation for modelling dynamics of
  biological systems
• Already successful in engineering of e.g.
  Internet, computer networks, goal of the Science
  of Global Ubiquitous Computer Challenge
• But new challenges in biology
• Why not simply increase the power or accuracy of
  techniques?
• Limited power of generating insight
• Novel scientific insight into biological
  processes
• Possible only through increase in understanding
  of fundamental principles
• Techniques aid in the process of scientific
  discovery (the crutch), but do not drive it
  (the torch)

6
Modelling

• Model is hypothesis about a biological process
  expressed in mathematical or computational
  formalism
• Does not seek to emulate the process
• Must produce outputs that are amenable to
  critical evaluation by biological experimentation
• Applicable to real biological data
• Must be falsifiable/verifiable by an experiment
• Multiplicity of formalisms needed ODE, SDE,
  process calculi

7
Modelling and Analysis

• Iterative cycle of
• Hypothesis forming, modelling, analysis
• Experimental validation, feedback
• Methods of analysis
• Simulation
• Automated verification, e.g. model checking
• Probabilistic verification
• Formal reasoning
• Key goals
• Realisation, when model consistently produces
  outputs that cannot be falsified by biological
  experiment
• In-silico prediction of organisms response
• Automation of the analysis process

8
Computer Science contribution

• Formal languages and models
• Principles and interaction primitives specific
  for biological processes
• Hybrid models for continuous and discrete
  dynamics
• Reasoning frameworks to establish important
  properties
• Control-theoretic techniques for robustness
• Automation of analysis research leading to the
  tools
• Efficient algorithms, for bioinformatics,
  analysis
• Quantitative and qualitative model checking
• Grid computing, e-Science
• Scalability
• Model reductions, abstraction
• Hierachical decomposition
• Compositionality

9
Verification via model checking

• Is it possible to reach a bistable state from the
  initial state?
• Does every execution from the initial state lead
  to bistable state?
• Properties expressible in temporal logic
• Models in process calculi
• Exhaustive search algorithms
• Can handle probability (PRISM model checker)
• Success with very large models

Entry point
10
Need for inter-disciplinarity

• Must ensure working together of
• Modellers, computer scientists, mathematicians,
  control engineers,
• Experimental biologists
• Appreciation of similarities
• State-transition models
• Reactions/interactions
• and differences
• Nature of the fundamental questions
• Uncertainty of experimental observation in
  biology
• Need for scientific rigour in all areas

11
What we can do

• Collaborate!
• Pathway Logic, BioSPI, BioCHAM, StateCharts,
  PRISM,
• e-Science, Integrative Biology,
• Enable collaboration, create inter-disciplinary
  centres
• Co-location of different disciplines, discipline
  hopping
• Ensure climate for industrial involvement
• Funding programmes
• Training programmes, communicate to public/media
• Numerate graduates, train in biology?
• Biologists, train in computational modelling?
• Aim for true discovery
• And not just for better instrumentation