Inferring nonlinear gene regulatory networks from gene expression data based on distance correlation

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• Gene regulatory networks (GRNs) exhibit complex nonlinear regulatory mechanisms.
• Accurate GRN reconstruction is crucial for understanding cellular systems.
• Existing methods may not fully capture nonlinear dependencies in gene expression data.

Purpose of the Study:

• To develop and evaluate novel algorithms for GRN inference using distance correlation (DC).
• To assess the performance of DC-based methods against mutual information (MI)-based approaches.
• To demonstrate the efficacy of DC in capturing nonlinear relationships within GRNs.

Main Methods:

• Incorporation of distance correlation (DC) for measuring nonlinear dependence in gene expression data.
• Development of three DC-based GRN inference algorithms: CLR-DC, MRNET-DC, and REL-DC.
• Comparative analysis using simulated benchmark GRNs (DREAM challenge, SynTReN) and an experimental E. coli SOS DNA repair network.

Main Results:

• DC-based algorithms demonstrated superior performance in GRN inference compared to MI-based methods.
• Receiver Operator Characteristic (ROC) and Precision-Recall (PR) curve analyses confirmed the outperformance.
• The proposed methods effectively identified regulatory relationships in both simulated and real biological networks.

Conclusions:

• Distance correlation is a powerful tool for inferring gene regulatory networks from expression data without distribution assumptions.
• DC-based algorithms offer significant advantages over MI-based methods for GRN reconstruction.
• This approach enhances the understanding of complex regulatory mechanisms in biological systems.