insulin-signaling-t2d

Dynamic notebooks for QSP usage case: Insulin Signaling in Type 2 Diabetes

Source repository

Notebooks quick links

• julia-jupiter.html,
• julia-pluto.html
• r-markdown.html
• r-jupiter.html

Abstract

Background: Currently the mathematical modeling is applied for drug discovery and development. The report preparation and presentation are time-demanding processes. Using the formats of dynamic reports like R Markdown, Jupiter or similar ones is a good decision because of the following:

1. Reducing time for report writing and update;
2. Easy results share, review, and reproduction;
3. Usage of interactivity capabilities, testing of “what-if” scenarios.

Objectives: Application and testing several software configurations for the development and analysis of dynamic reports. Testing the integration of the Heta-based platform into the presented environments. Comparison of the capabilities and technical issues for different configurations.

Methods: We tested several configurations for dynamic reporting:

• Julia environment + HetaSimulator.jl + Jupiter notebooks.
• Julia environment + HetaSimulator.jl + Pluto.jl;
• R environment + heta-compiler + mrgsolve + R Markdown;
• R environment + heta-compiler + mrgsolve + Jupiter notebook;

Results: The dynamic report was created for each model and configuration. It included model code loading, single and Monte-Carlo simulations, and visualization plots. All settings and configuration files are shared on GitHub.

Discussion: The following features of modeling infrastructure is very important for successful and effective dynamic reports:

1. representation of QSP model in human-readable format or another unified format,
2. loading of a model from programming environment,
3. Simulation engine available from programming environments, like R or Julia. If you use a standardized modeling environment like the Heta-based modeling platform and heta-compiler tool the development of a dynamic report requires no more than one hour.

Project content

• /julia : files to run in plain julia, scenarios tables for all notebooks
• /julia-jupiter : The Jupiter-based notebook, can be run with using IJulia; notebook(dir=".") in Julia
• /julia-pluto : The Pluto-based notebook, can be run with using Pluto; Pluto.run() in Julia
• /r-markdown : RMarkdown notebook, can be run from RStudio using knit mechanism
• /r-jupiter : The Jupiter-based notebook, can be run with anaconda. Commands: conda activate snowflakes, jupyter notebook --notebook-dir=Y:/PLATFORMS/insulin-signaling-t2d/r-jupiter)
• /static : the result of the notebooks execution, not interactive: julia-jupiter.html, julia-pluto.html, r-markdown.html, r-jupiter.html
• /src : source files of QSP model, Heta-based format + downloaded SBML
• platform.json : Heta platform's declaration file
• README.md : this file

Original model

The model and data were reconstructed from the article:

Brannmark C, Nyman E, Fagerholm S, Bergenholm L, Ekstrand EM, Cedersund G, Stralfors P. Insulin Signaling in Type 2 Diabetes: Experimental and modeling analysis reveal mechanisms of insulin resistance in human adipocytes. Journal of biological chemistry. 2013 288(14):9867–9880. DOI: 10.1074/jbc.M112.432062

The SBML version was downloaded from BioModels https://www.ebi.ac.uk/biomodels/BIOMD0000000448

Contributors

• @metelkin

The model and data in the study were reproduced from the published study. The authors of the original study are: Brannmark C, Nyman E, Fagerholm S, Bergenholm L, Ekstrand EM, Cedersund G, Stralfors P.