Lecture 8

Question 1

Modelling in Systems Biology

Answer 1

1) Hypothesis
2) Experiment
3) Data
4) Modelling
5) Simulation

• All models are wrong, but some are useful
• Models are always abstractions
  -> simplifications that help understanding principles of regulation
• modeling involve iteration cycles
  -> fit model predictions with experimental data - but how?

Question 2

Bootstrapping Example

Answer 2

• EMT in Cancer => FOXA1 pioneering transcription factor => repressed in EMT
• Preferential Binding to Epithelial/Mesenchymal genes?
• Found 481 peaks in 100 genes
• 66000 peaks altogether
• A lot or a little binding? => use bootstrapping (10000x 100 genes selected)
• Signi!cant binding to epithelial/mesenchymal genes

Question 3

Cross Validation

Answer 3

• There is a dierence between model !tting & prediction
• Can !t model to training data, but poor prediction => over!tting
• Split data into training & test data
• Average prediction: goodness for model
• Use Likelihood ratio model selection -> exercises

Question 4

Genetic Algorithm

Answer 4

• Directed search algorithm based on mechanics of biological evolution
• reproduction -> modification -> evaluation -> population
• Population
• Chromosomes could be:
  -> Bit strings (0101 … 1100)
  -> Real numbers (43.2 -33.1 … 0.0 89.2)
  -> … any data structure …
• Reproduction
  -> Parents are selected at random with selection chances biased in relation to chromosome evaluations.
• Modi!cation
  -> Mutation => local adaptation
  -> Crossing over
  => accelerates search in early evolution of population
• Evaluation
  -> Score !tness of individual via objective function
  -> Only link between GA & problem it solves

Question 5

Traveling Salesman Problem

Answer 5

• Given a list of cities and the distances between each pair of cities, what is the shortest possible route that visits each city exactly once and returns to the origin city?
• Complex problem, if number of cities large
• Formulate as GA problem
  => Chromosome: ordered list of cities
  1) London 2) Venice
  3) Berlin 5) Beijing 7) Tokyo
  4) Singapore 6) Phoenix 8) Sydney
• Objective function: sum of distances between ordered list elements

Question 6

Model Simplification

Answer 6

• Works on the model topology to reduce number of variables & parameters
  a) Omit elements
  b) Fix elements
  c) Simplify formulas
  d) Lump elements
  e) Dynamic black box model
  f) Global flux modes

Question 7

Spatio-Temporal Organization of Life

Answer 7

• Chemical reactions => Mathematical modeling
• Attractors, Bifurcations, Multi-stability, Macroscopic organization * Stability & Plasticity of complex systems
• Adiabatic elimination & Slaving Principle applicable on multiple time scales
  -> “Commanding Process” depends on time/space scale of observation

Why is Life? Because Life can!