GREMA

GREMA infers gene regulatory networks (GRNs) and produces emulated GRNs (eGRNs) with confidence levels to enable robust modeling of gene interactions using evolutionary optimization.

Key Features:

• Non-linear ODE Models: GREMA uses non-linear ordinary differential equation models, specifically S-system and Hill function-based models, to represent dynamic gene interactions.
• Evolutionary Modelling Algorithm (EMA): GREMA employs an Evolutionary Modelling Algorithm that leverages evolutionary intelligence, including crowd wisdom and evolutionary strategies, to address large-scale parameter optimization in underdetermined eGRN models.
• Intelligent Genetic Algorithm: The EMA implements an intelligent genetic algorithm to optimize model parameters efficiently despite underdetermination.
• Confidence Levels: GREMA assigns a confidence level to each inferred regulation and ranks inferred regulations in descending order of confidence.
• Performance and Validation: GREMA reported a mean accuracy improvement of 19.2% using S-system models on benchmark datasets such as DREAM4 and SOS DNA repair.
• Comparison with Existing Methods: GREMA has been tested against existing methods on similar datasets and demonstrated superior accuracy and robustness of inferred regulations.

Scientific Applications:

• Systems Biology: Reconstruction of gene regulatory networks for systems-level analysis of biological processes.
• Bioinformatics: Computational inference of GRNs for data-driven bioinformatics studies.
• Genetic Engineering: Identification of regulatory interactions relevant to genetic engineering applications.
• Personalized Medicine: Informing models of gene regulation that can support personalized medicine investigations.
• Synthetic Biology: Modeling and design of gene regulatory circuits for synthetic biology.

Methodology:

GREMA fits non-linear ODE models (S-system and Hill functions) using an Evolutionary Modelling Algorithm (EMA) that incorporates crowd wisdom and evolutionary strategies together with an intelligent genetic algorithm to optimize parameters, assigns confidence levels to inferred regulations, and orders regulations by descending confidence.