CoChat

Dynamical modeling and multi-experiment fitting with PottersWheel

Open access
Maiwald, Thomas, Timmer, Jens · 2008 · Bioinformatics

The DOI matches its Crossref record; 26 of the 27 references it cites were found in a registry or an index.

Abstract

Motivation: Modelers in Systems Biology need a flexible framework that allows them to easily create new dynamic models, investigate their properties and fit several experimental datasets simultaneously. Multi-experiment-fitting is a powerful approach to estimate parameter values, to check the validity of a given model, and to discriminate competing model hypotheses. It requires high-performance integration of ordinary differential equations and robust optimization. Results: We here present the comprehensive modeling framework Potters-Wheel (PW) including novel functionalities to satisfy these requirements with strong emphasis on the inverse problem, i.e. data-based modeling of partially observed and noisy systems like signal transduction pathways and metabolic networks. PW is designed as a MATLAB toolbox and includes numerous user interfaces. Deterministic and stochastic optimization routines are combined by fitting in logarithmic parameter space allowing for robust parameter calibration. Model investigation includes statistical tests for model-data-compliance, model discrimination, identifiability analysis and calculation of Hessian- and Monte-Carlo-based parameter confidence limits. A rich application programming interface is available for customization within own MATLAB code. Within an extensive performance analysis, we identified and significantly improved an integrator–optimizer pair which decreases the fitting duration for a realistic benchmark model by a factor over 3000 compared to MATLAB with optimization toolbox. Availability: PottersWheel is freely available for academic usage at http://www.PottersWheel.de/. The website contains a detailed documentation and introductory videos. The program has been intensively used since 2005 on Windows, Linux and Macintosh computers and does not require special MATLAB toolboxes. Contact: maiwald@fdm.uni-freiburg.de Supplementary information: Supplementary data are available at Bioinformatics online.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/bioinformatics/btn350

Cites

26 of 26 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1. PW workflow as described in Section 2.1.
Fig. 1. PW workflow as described in Section 2.1.CC BY-NC · PubMed Central
Fig. 2. PW Equalizer. The PW Equalizer comprises 10 pairs of sliders. Each pair can be attached to one of the fitted parameters in the list box below the slider pair. The left slider is used for fine-tuning and the right slider changes the magnitude of the selected parameter. Text boxes on the one hand reflect the current parameter value and can on the other hand be used directly to specify a certain value. Thirty-three buttons and combo boxes provide direct access to the most important functions concerning multi-experiment fitting and analysis.
Fig. 2. PW Equalizer. The PW Equalizer comprises 10 pairs of sliders. Each pair can be attached to one of the fitted parameters in the list box below the slider pair. The left slider is used for fine-tuning and the right slider changes the magnitude of the selected parameter. Text boxes on the one hand reflect the current parameter value and can on the other hand be used directly to specify a certain value. Thirty-three buttons and combo boxes provide direct access to the most important functions concerning multi-experiment fitting and analysis.CC BY-NC · PubMed Central
Fig. 3. Integration time and accuracy. Left panel: the mean integration time of all 13 supported integrators with specified relative and absolute tolerance (Trel, Tabs) is displayed. The integrator either includes the ODE (Rows 1 and 2) or is attached to an ODE compiled as C MEX file or saved as a normal MATLAB function. The reference is the MATLAB integrator ODE15s for stiff systems with a MATLAB ODE (last row, integration time 0.14 s, deviation 2×10−4). DOPRI5 for non-stiff systems is 919 times faster if the ODE is included (Row 1) and 26 times faster using a MEX ODE (Row 13). RADAU5, a stiff integrator, is 162 times faster with included ODE (Row 2) and 11 times faster with a MEX ODE (Row 3) compared to the reference. Increasing RADAU integration accuracy by 10−5 doubles the integration time (Rows 5–10). Right panel: integration with RADAU using high tolerances of 10−8 and 10−11 serves here as a gold standard to estimate the accuracy of all integrators by quantifying the mean deviation between the calculated trajectories. RADAU5 (tolerances 10−3 and 10−6, Row 2) is not only faster than ODE15s, but also five times more accurate.
Fig. 3. Integration time and accuracy. Left panel: the mean integration time of all 13 supported integrators with specified relative and absolute tolerance (Trel, Tabs) is displayed. The integrator either includes the ODE (Rows 1 and 2) or is attached to an ODE compiled as C MEX file or saved as a normal MATLAB function. The reference is the MATLAB integrator ODE15s for stiff systems with a MATLAB ODE (last row, integration time 0.14 s, deviation 2×10−4). DOPRI5 for non-stiff systems is 919 times faster if the ODE is included (Row 1) and 26 times faster using a MEX ODE (Row 13). RADAU5, a stiff integrator, is 162 times faster with included ODE (Row 2) and 11 times faster with a MEX ODE (Row 3) compared to the reference. Increasing RADAU integration accuracy by 10−5 doubles the integration time (Rows 5–10). Right panel: integration with RADAU using high tolerances of 10−8 and 10−11 serves here as a gold standard to estimate the accuracy of all integrators by quantifying the mean deviation between the calculated trajectories. RADAU5 (tolerances 10−3 and 10−6, Row 2) is not only faster than ODE15s, but also five times more accurate.CC BY-NC · PubMed Central
Fig. 4. Optimization performance of Levenberg-Marquardt. (A): Deviation compared to true parameters. The mean deviation over all parameters (dashed red) reaches 1.0994 after around 240 iterations. After the final iteration, the maximum deviation was 1.30108, indicating that no parameter value differed more than 30% from the true value. (B): Parameter values during fitting. (C): χ2-value during fitting and fit settings.
Fig. 4. Optimization performance of Levenberg-Marquardt. (A): Deviation compared to true parameters. The mean deviation over all parameters (dashed red) reaches 1.0994 after around 240 iterations. After the final iteration, the maximum deviation was 1.30108, indicating that no parameter value differed more than 30% from the true value. (B): Parameter values during fitting. (C): χ2-value during fitting and fit settings.CC BY-NC · PubMed Central
Fig. 5. Application of multi-experiment fitting. If only one experiment with continuous stimulations for each driving input player cyto-c and SMAC (A), four observables, 10% rel.+10% absolute error, AU: arbitrary units) is fitted 200 times with varying initial guess for the parameters, the distributions of the calibrated parameter values is rather broad: the parameters are not identifiable (B). Combination of all four experiments (C) leads to significantly narrowed parameter distribution decreasing the relative uncertainty from over 100% to below 2% for 11 of 17 parameters (D). Note that the distribution represented by the horizontal bars should not be mistaken for a confidence interval (see Section 2.6). Therefore, true values (red stars) may lie outside of the parameter distributions.
Fig. 5. Application of multi-experiment fitting. If only one experiment with continuous stimulations for each driving input player cyto-c and SMAC (A), four observables, 10% rel.+10% absolute error, AU: arbitrary units) is fitted 200 times with varying initial guess for the parameters, the distributions of the calibrated parameter values is rather broad: the parameters are not identifiable (B). Combination of all four experiments (C) leads to significantly narrowed parameter distribution decreasing the relative uncertainty from over 100% to below 2% for 11 of 17 parameters (D). Note that the distribution represented by the horizontal bars should not be mistaken for a confidence interval (see Section 2.6). Therefore, true values (red stars) may lie outside of the parameter distributions.CC BY-NC · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 1

  1. Tarantola, 2004, Inverse Problem Theory and Model Parameter Estimation.

Cited by

See all 231

Metadata

Authors
Maiwald, T., & Timmer, J.
Affiliations
University of Freiburg (DE)
Corresponding author
Maiwald, Thomas
Venue
Bioinformatics
Published
2008-07-09
Volume
24
Issue
18
Pages
2037-2043
ISSN
  • 1367-4803 (ISSN-L)
  • 1367-4811
OpenAlex
W2166071821
Cited by
231
References
27
Citations by year
2026: 1 · 2025: 1 · 2024: 2 · 2023: 4 · 2022: 8 · 2021: 25 · 2020: 7 · 2019: 10 · 2018: 9 · 2017: 13 · 2016: 14 · 2015: 16 · 2014: 17 · 2013: 19 · 2012: 30
FWCI
7.55
Type
journal article
Language
en
Open access
Locations
  • Bioinformatics (publishedVersion, cc-by-nc)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by-nc)
  • PubMed Central (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
Funders
  • Bundesministerium für Bildung und Forschung (0313074D)
  • European Social Fund
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene Regulatory Network Analysis
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Microbial Metabolic Engineering and Bioproduction
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Bioinformatics and Genomic Networks
Keywords
dynamic modeling, ordinary differential equations, parameter estimation, inverse problems, model discrimination, identifiability analysis, stochastic optimization, signaling pathways, metabolic networks, MATLAB toolbox
MeSH terms
Computer Simulation, Programming Languages, Software, Signal Transduction, Systems Biology

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.