GRNMOPT: Inference of gene regulatory networks based on a multi-objective optimization approach

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• Gene regulatory network (GRN) reconstruction is crucial for understanding biological processes.
• Nonlinear ordinary differential equations (ODEs) are effective for GRN prediction.
• Systematic determination of decay rates and time delays in ODEs models for GRNs is underexplored.

Purpose of the Study:

• To develop a comprehensive optimization framework for accurate GRN inference.
• To systematically estimate key parameters like decay rate and time delay in ODEs models.
• To improve the accuracy of GRN reconstruction.

Main Methods:

• Introduced GRNMOPT, a novel methodology for inferring GRNs from time-series and steady-state data.
• Employed ODEs models incorporating decay rate and time delay for authentic gene regulation representation.
• Utilized a multi-objective optimization approach to concurrently optimize decay rate and time delay, deriving Pareto optimal sets.
• Applied XGBoost for feature importance calculation to identify potential regulatory gene links.

Main Results:

• GRNMOPT demonstrated commendable performance across various network scales on simulated (DREAM4) and real gene expression datasets (Yeast, IRMA, E. coli).
• Cross-validation experiments confirmed the robustness of the GRNMOPT methodology.
• The approach effectively improved accuracy metrics such as AUROC and AUPR.

Conclusions:

• Proposed GRNMOPT, a novel approach for GRN inference utilizing a multi-objective optimization framework.
• GRNMOPT significantly enhances GRN inference accuracy.
• This method provides a powerful tool for advancing GRN research.