Global sensitivity analysis in stochastic simulators of uncertain reaction networks
M Navarro Jimenez1, O P Le Maître1, O M Knio1
1CEMSE Division, King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia.
This study introduces a novel global sensitivity analysis for chemical systems, accounting for both parameter uncertainties and inherent stochasticity. The method quantifies variability contributions from parameters and reaction channels, offering a more comprehensive understanding of model behavior.
Area of Science:
- Computational Chemistry
- Chemical Kinetics
- Systems Biology
Background:
- Stochastic chemical models face uncertainties from kinetic parameters and inherent randomness.
- Current uncertainty quantification relies on sensitivity analyses of statistical moments.
- Existing methods often overlook the interplay between parametric and inherent variability.
Purpose of the Study:
- To develop an original global sensitivity analysis method for stochastic chemical systems.
- To simultaneously address uncertainties in kinetic parameters and inherent stochastic dynamics.
- To provide a unified framework for quantifying multiple sources of variability.
Main Methods:
- Utilizing Sobol's decomposition to partition variance contributions.
- Treating inherent stochasticity via Poisson processes in a random-time-change representation.
- Developing a sampling algorithm for estimating partial variances and sensitivity indices.
Main Results:
- The proposed method effectively decomposes variance into contributions from parameter subsets and stochastic reaction channels.
- Demonstrated application on birth-death and Schlögl models showcases the method's utility.
- Identified and quantified interactions between parametric and inherent variability sources.
Conclusions:
- The novel global sensitivity analysis offers a richer understanding of variability in stochastic chemical models.
- This approach provides a more complete picture than traditional local sensitivity analyses.
- The method is crucial for accurate uncertainty quantification in complex chemical systems.
More Related Videos
13:34A Combined 3D Tissue Engineered In Vitro/In Silico Lung Tumor Model for Predicting Drug Effectiveness in Specific Mutational Backgrounds
Published on: April 6, 2016
05:57Author Spotlight: In Silico Creation and Impact of Carbonylated Amino Acids on Protein Structure and Function
Published on: April 26, 2024
Related Concept Videos
Reaction Mechanisms: Rate-limiting Step Approximation
Propagation of Uncertainty from Systematic Error
Quantitative Aspects of Drug-Receptor Interaction
Global Regulatory Systems
Predicting Reaction Outcomes
Mechanistic Models: Compartment Models in Individual and Population Analysis
