AMICI

AMICI provides efficient simulation and sensitivity analysis for ordinary and differential-algebraic equation models in systems biology, enabling precise gradient computation for parameter estimation and uncertainty quantification.

Key Features:

• Multi-Language Support: Interfaces with Python, C++, and MATLAB and reads and compiles models into MATLAB .mex files, C++ executables, and Python modules.
• SUNDIALS Solver Integration: Integrates with the SUNDIALS solvers CVODES for ordinary differential equations and IDAS for algebraic differential equations.
• Model Format Support: Handles biochemical reaction models described in SBML and PySB formats.
• Efficient Simulation Routines: Provides optimized, scalable simulation routines suitable for large and complex biochemical reaction models.
• Advanced Sensitivity Analysis: Supports forward sensitivity analysis, steady-state sensitivity analysis, and adjoint sensitivity analysis for precise gradient computation.
• Gradient-Based Parameter Estimation: Enables computation of gradients required for likelihood-based output functions and differential-equation-constrained optimization.
• Scalable and Modular Design: Implements a modular architecture that supports scalable operations and customization for models of varying complexity.

Scientific Applications:

• Systems Biology Modeling: Simulation and analysis of cellular signal transduction and other biochemical reaction networks using ODE and DAE models.
• Parameter Estimation: Gradient-based parameter estimation and optimization for biochemical reaction models.
• Uncertainty Quantification: Sensitivity-based assessment of parameter uncertainty and its impact on model outputs.
• Model Behavior Exploration: Exploration of model responses and steady-state properties under varying conditions using sensitivity analyses.

Methodology:

Reads and compiles SBML/PySB models into MATLAB .mex, C++ executables, or Python modules; integrates with SUNDIALS CVODES and IDAS; performs forward, steady-state, and adjoint sensitivity analyses; and computes gradients for likelihood-based parameter estimation.