HiSCF: leveraging higher-order structures for clustering analysis in biological networks

Area of Science:

• Systems Biology
• Bioinformatics
• Network Science

Background:

• Biological network analysis aims to group entities into functional modules for system understanding.
• Current clustering methods often rely on lower-order connectivity, neglecting higher-order network motifs.

Purpose of the Study:

• To present HiSCF, a novel clustering framework utilizing higher-order network structures.
• To improve the accuracy of functional module identification in biological networks.

Main Methods:

• Developed a clustering framework named HiSCF.
• Employed higher-order Markov stochastic processes to exploit network motifs for clustering.
• Evaluated performance against state-of-the-art clustering models.

Main Results:

• HiSCF achieved superior performance in protein complex identification and gene co-expression module detection.
• Demonstrated the value of higher-order network motifs for biological network analysis.
• Highlighted HiSCF's capability in identifying overlapping protein complexes and inferring signaling pathways.

Conclusions:

• Incorporating higher-order network motifs provides novel insights into biological network organization.
• HiSCF offers a powerful approach for uncovering complex biological functional modules.
• The framework reveals rich higher-order organizational structures within biological networks.