A global sensitivity analysis approach for morphogenesis models. Issue 1 (December 2015)

• Record Type: Journal Article
• Authors: Boas, Sonja; Navarro Jimenez, Maria; Merks, Roeland; Blom, Joke
• Is Part Of: BMC systems biology. Volume 9:Issue 1(2015)
• Page Start: 1
• Page End: 15
• Publication Date: 2015-12
• Subjects: Morphogenesis -- Vascular network development -- Computational modeling -- Cellular Potts model -- Global sensitivity analysis -- Sobol' indices -- Polynomial chaos expansion; Biological systems -- Periodicals; Biology -- Research -- Periodicals; Cell physiology -- Periodicals; Genes -- Analysis -- Periodicals; 571
• DOI: 10.1186/s12918-015-0222-7
• Languages: English
• ISSNs: 1752-0509

Abstract

Abstract Background Morphogenesis is a developmental process in which cells organize into shapes and patterns. Complex, non-linear and multi-factorial models with images as output are commonly used to study morphogenesis. It is difficult to understand the relation between the uncertainty in the input and the output of such 'black-box' models, giving rise to the need for sensitivity analysis tools. In this paper, we introduce a workflow for a global sensitivity analysis approach to study the impact of single parameters and the interactions between them on the output of morphogenesis models. Results To demonstrate the workflow, we used a published, well-studied model of vascular morphogenesis. The parameters of this cellular Potts model (CPM) represent cell properties and behaviors that drive the mechanisms of angiogenic sprouting. The global sensitivity analysis correctly identified the dominant parameters in the model, consistent with previous studies. Additionally, the analysis provided information on the relative impact of single parameters and of interactions between them. This is very relevant because interactions of parameters impede the experimental verification of the predicted effect of single parameters. The parameter interactions, although of low impact, provided also new insights in the mechanisms ofin silico sprouting. Finally, the analysis indicated that the model could be reduced by one parameter. Conclusions We propose global sensitivity analysis as an alternative approach to study the mechanisms of morphogenesis. Comparison of the ranking of the impact of the model parameters to knowledge derived from experimental data and from manipulation experiments can help to falsify models and to find the operand mechanisms in morphogenesis. The workflow is applicable to all 'black-box' models, including high-throughputin vitro models in which output measures are affected by a set of experimental perturbations.