T08 - Statistics and Numerics for Dynamical Modeling

Clemens Kreutz*, Bernhard Steiert
* This email address is being protected from spambots. You need JavaScript enabled to view it.

 

Description

A successful mathematical description of cell biological processes based on experimental data requires efficient and reliable numerical methods for parameter estimation as well as a suitable statistical methodology to reconstruct the underlying biochemical reaction networks. In this tutorial, statistical and numerical aspects for dynamic modeling in Systems Biology are discussed and the computational implementation is demonstrated. One major focus is the assessment of uncertainties of both, parameters and model predictions, which is efficiently and intuitively judged by the profile likelihood.

In summary, the following aspects are discussed:

  • appropriate numerical algorithms for parameter estimation [1]
  • judging the quality of experimental data
  • model discrimination by the likelihood ratio tests
  • identifiability analysis and confidence intervals for estimated parameters [2]
  • observability analysis and confidence intervals for model predictions [3]
  • selection of informative new experimental conditions [4]

In the tutorial, we use a basic system of ordinary differential equations (ODEs) to implement and illustrate the methods in MATLAB together with the participants. In addition, a published model for JAK-STAT signaling [5] and a comprehensive software package for quantitative dynamic modelling [1] is introduced. The illustrated methodology has been awarded three times as “Best Performer” in parameter estimation challenges within the Dialogue for Reverse Engineering Assessment and Methods (DREAM) competitions [6-8].

 

References

[1] Raue A, Schilling M, Bachmann J, Matteson A, Schelker M, Kaschek D, Hug S, Kreutz C, Harms BD, Theis F, Klingmüller U and Timmer J.: Lessons Learned from Quantitative Dynamical Modeling in Systems Biology. PLoS ONE 8(9), e74335, 2013
[2] Raue A, Kreutz C, Maiwald T, Bachmann J, Schilling M, Klingmüller U, Timmer J.: Structural and practical identifiability analysis of partially observed dynamical models by exploiting the profile likelihood. Bioinformatics. 2009 Aug 1;25(15):1923-9.
[3] Kreutz C, Raue A, Timmer J: Likelihood based observability analysis and confidence intervals for predictions of dynamic models. BMC Systems Biology 2012; 6.
[4] Steiert B, Raue A, Timmer J, Kreutz C: Experimental design for parameter estimation of gene regulatory networks. PLoS ONE 2012; 7, e40052.
[5] Becker V, Schilling M, Bachmann J, Baumann U, Raue A, Maiwald T, Timmer J and Klingmueller U, Covering a broad dynamic range: information processing at the erythropoietin receptor, Science 328(5984), 1404-1408, 2010.
[6] DREAM6 – Estimation of Model Parameters Challenge, 2011. http://www.the-dream-project.org/challenges/dream6-estimation-model-parameters-challenge
[7] DREAM7 – Network Topology and Parameter Inference Challenge, 2012. http://www.the-dream-project.org/challenges/network-topology-and-parameter-inference-challenge
[8] DREAM8 – Whole Cell parameter estimation challenge, 2013. https://www.synapse.org/#!Synapse:syn1876068 

 

Provisional schedule

Date: September 7th, 2014.

9:00 – 10:00 Introduction and Motivation, some applications of dynamic modelling based on experimental data are presented.
10:00 – 10:30 Coffee Break
10:30 – 11:00 A basic example and computational implementation of parameter estimation and calculation of the profile likelihood to assess parameter uncertainties and detect non-identifiabilities.
11:00 – 12:30 Hands on session 1:
- Simulation of data for the basic example and a slightly changed setting
- Parameter estimation, Profile likelihood
12:30 – 13:30 Lunch
13:30 – 15:00 Presentation of the comprehensive methodology and its introduction of its implementation in the publicly available “Data 2 Dynamics Software”. https://bitbucket.org/d2d-development/d2d-software/wiki/Home
15:00 - 16:00 Hand on session 2:
Analysis of real experimental data using the “Data 2 Dynamics Software”
16:00 – 17:00 Presentation of other methodological aspects and recently published approaches. Discussion and Summary.

 

Technical requirements and intended audience

For parts of the tutorial the D2D data2dynamics software package is used which runs under MATLAB. To be able to perform the hand-on part of the tutorial, the participants should bring their own computers. WLAN is recommended to be able to distribute code between the participants. MATLAB (best R2010 or newer) including the Symbolic and Optimization Toolbox is required. Further information concerning the system requirements as well as for the software itself are found at https://bitbucket.org/d2d-development/d2d-software/wiki/Installation

The tutorial is given for scientists with a theoretical background (bioinformatics, physics, theoretical biology, statistics) or for biologists working in the Systems Biology field.
The participants should have at least an educational level of a master- or PhD student. The room is limited to a maximum of 45 people.

 

Instructors

The tutorial is given by Bernhard Steiert and Clemens Kreutz. They have long-term experience in data-based modelling and published several new statistical approaches for modelling in Systems Biology as well as applications with experimental collaborators.

Bernhard Steiert and Clemens Kreutz participated three times in the DREAM (Dialogue for Reverse Engineering Assessment and Methods) parameter estimation challenges and were awarded as best performing every time.

Latest News

Awards ECCB'14 awards have been announced during the closing ceremony of ECCB'14 on Wednesday September... Read more
Next conferences: ISMB/ECCB & JOBIM 2015 Next ECCB will be held in conjunction with ISMB in Dublin, Ireland, July 10-14, 2015: ISMB/ECCB... Read more

Silver Sponsors

Other Exhibitors