NSRGRN: a network structure refinement method for gene regulatory network inference.

Area of Science:

• Systems Biology
• Computational Biology
• Genomics

Background:

• Gene regulatory network (GRN) inference is vital for understanding disease mechanisms.
• Identifying redundant regulations in GRNs remains a significant computational challenge.
• Existing methods struggle to balance topological properties and edge importance measures.

Purpose of the Study:

• To develop a Network Structure Refinement method for GRN (NSRGRN) inference.
• To effectively combine topological properties and edge importance measures for improved GRN inference.
• To address the limitations of current methods in identifying and reducing redundant gene regulations.

Main Methods:

• NSRGRN employs a two-part approach: preliminary ranking and a novel network structure refinement (NSR) algorithm.
• The NSR algorithm optimizes local topology using Conditional Mutual Information with Directionality and network motifs.
• Global topology is maintained by balancing local optimization with lower and upper network structures.

Main Results:

• NSRGRN demonstrated superior all-round performance compared to six state-of-the-art methods across 26 datasets.
• The NSR algorithm, as a post-processing step, enhanced the results of other GRN inference methods.
• The proposed method effectively identifies and reduces redundant regulations in gene regulatory networks.

Conclusions:

• NSRGRN offers a robust and effective solution for GRN inference, addressing key challenges in the field.
• The NSR algorithm provides a valuable tool for improving the accuracy and reliability of computational GRN analysis.
• This work contributes to advancing systems biology by providing a more precise understanding of gene regulation.