Moment-based parameter inference with error guarantees for stochastic reaction networks.

Area of Science:

• Systems Biology
• Computational Biology
• Biochemical Kinetics

Background:

• Parameter inference for stochastic reaction networks is hindered by intractable likelihood functions and limitations in current error quantification methods.
• Existing approaches lack theoretical guarantees when likelihood approximations are necessary, especially for non-linear models and single-cell data.

Purpose of the Study:

• To develop a robust method for inferring parameters of biochemical kinetic models from single-cell data, addressing uncertainty quantification challenges.
• To provide theoretical guarantees for parameter bounds in stochastic reaction networks, applicable to both steady-state and time-resolved data.

Main Methods:

• Proposed a convex optimization approach utilizing moment equations and moment matrices derived from observational data.
• Constructed moment intervals from observations to constrain parameters within convex sets, ensuring bounds contain true parameters.
• Avoided direct computation of likelihoods or simulation of the forward problem, circumventing computational intractability.

Main Results:

• Successfully provided uncertainty quantification and error guarantees for parameter inference in stochastic reaction networks.
• Demonstrated applicability to non-linear and rational propensity networks, including the Schlögl model and toggle switch models.
• Validated the method's effectiveness for data integration and prediction of latent species statistics using synthetic data.

Conclusions:

• The proposed method offers a computationally efficient and theoretically sound framework for parameter inference and uncertainty quantification in biochemical models.
• This approach overcomes limitations of traditional methods by directly using data moments without requiring likelihood approximations or simulations.
• The technique is versatile, applicable to various biochemical models and data types, advancing the analysis of complex biological systems.