1 / 98

Inferring gene regulatory networks from transcriptomic profiles

Dirk Husmeier. Inferring gene regulatory networks from transcriptomic profiles. Biomathematics & Statistics Scotland. Overview. Introduction Limitations Methodology Application to morphogenesis Application to synthetic biology. Objective: reverse engineering regulatory networks.

dolph
Download Presentation

Inferring gene regulatory networks from transcriptomic profiles

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Dirk Husmeier Inferring gene regulatory networks from transcriptomic profiles Biomathematics & Statistics Scotland

  2. Overview • Introduction • Limitations • Methodology • Application to morphogenesis • Application to synthetic biology

  3. Objective: reverse engineering regulatory networks From Sachs et al Science 2005

  4. Network reconstruction from postgenomic data

  5. Model Parameters q Probability theory  Likelihood

  6. Mechanistic model: description with differential equations Concentrations Kinetic parameters q Rates

  7. Model Parameters q Probability theory  Likelihood

  8. 1) Practical problem: numerical optimization q 2) Conceptual problem: overfitting ML estimate increases on increasing the network complexity

  9. Overfitting problem True pathway Poorer fit to the data Equal or better fit to the data Poorer fit to the data

  10. Regularization E.g.: Bayesian information criterion (BIC) Regularization term Data misfit term Maximum likelihood parameters Number of parameters Number of data points

  11. Likelihood BIC Complexity Complexity

  12. Model selection: find the best pathway Select the model with the highest posterior probability: This requires an integration over the whole parameter space:

  13. MCMC based schemes q Problem: excessive computational costs

  14. Accuracy Mechanistic models DynamicBayesian networks Computational complexity

  15. Marriage between graph theory and probability theory Friedman et al. (2000), J. Comp. Biol. 7, 601-620

  16. Bayes net ODE model

  17. Model Parameters q Bayesian networks: integral analytically tractable!

  18. UAI 1994

  19. Example: 2 genes 16 different network structures Compute

  20. Identify the best network structure Ideal scenario: Large data sets, low noise

  21. Uncertainty about the best network structure Limited number of experimental replications, high noise

  22. Sample of high-scoring networks

  23. Sample of high-scoring networks Feature extraction, e.g. marginal posterior probabilities of the edges

  24. Sample of high-scoring networks Feature extraction, e.g. marginal posterior probabilities of the edges Uncertainty about edges High-confident edge High-confident non-edge

  25. Sampling with MCMC Number of structures Number of nodes

  26. Overview • Introduction • Limitations • Methodology • Application to morphogenesis • Application to synthetic biology

  27. Model Parameters q Bayesian networks: integral analytically tractable!

  28. Homogeneity assumption Parameters don’t change with time

  29. Homogeneity assumption Parameters don’t change with time

  30. Limitations of the homogeneity assumption

  31. Overview • Introduction • Limitations • Methodology • Application to morphogenesis • Application to synthetic biology

  32. Example: 4 genes, 10 time points

  33. Standard dynamic Bayesian network: homogeneous model

  34. Limitations of the homogeneity assumption

  35. Our new model: heterogeneous dynamic Bayesian network. Here: 2 components

  36. Changepoint model Parameters can change with time

  37. Changepoint model Parameters can change with time

  38. Our new model: heterogeneous dynamic Bayesian network. Here: 2 components

  39. Our new model: heterogeneous dynamic Bayesian network. Here: 3 components

  40. Extension of the model q

  41. Extension of the model q Allocation vector h k Number of components (here: 3)

  42. Analytically integrate out the parameters q Allocation vector h k Number of components (here: 3)

  43. RJMCMC within Gibbs P(network structure | changepoints, data) P(changepoints | network structure, data) Birth, death, and relocation moves

  44. Model extension So far:non-stationarity in the regulatory process

  45. Non-stationarity in the network structure

  46. Flexible network structure .

  47. Model Parameters q

  48. Model Parameters q Use prior knowledge!

More Related