Complex Networks in Bioactive Peptide Research: A Methodological Review

Area of Science:

• Computational chemistry and bioinformatics
• Peptide science and drug discovery

Background:

• Bioactive peptides are a diverse and therapeutically important class of molecules.
• Exploring peptide chemical space is challenging due to its vastness, heterogeneity, and fragmented annotation.
• Complex networks provide a computational framework for organizing and analyzing peptide diversity.

Purpose of the Study:

• To review graph-based peptide informatics methodologies for exploring bioactive peptide chemical space.
• To examine network construction, topological analysis, and similarity searching models.
• To highlight computational resources for peptide network science.

Main Methods:

• Data integration and curated repositories for peptide information.
• Descriptor-based representations and similarity-driven network construction (Chemical Space Networks, Half-Space Proximal Networks, Metadata Networks).
• Topological analysis (threshold selection, community detection, centrality) and Multi-query Similarity Searching Models.

Main Results:

• Peptide sequences can be projected into multidimensional spaces for network construction.
• Topological analysis aids in identifying representative peptides and scaffolds.
• Development of training-independent similarity searching models as alternatives to supervised predictors.

Conclusions:

• Complex networks offer a mature and interpretable paradigm for structured exploration of bioactive peptides.
• Computational resources like StarPepDB, StarPep Toolbox, and StarPepWeb facilitate accessible and reproducible peptide network science.
• This approach enhances the systematic analysis and discovery of novel bioactive peptides.