GHOST

GHOST performs global alignment of biological networks, specifically protein interaction networks, by establishing correspondences between proteins while preserving interactions and incorporating homologous evidence.

Key Features:

• Novel Spectral Signature: GHOST computes a spectral signature to assess topological similarity among subnetworks and guide node correspondences across networks.
• Seed-and-Extend Alignment with Local Search: The alignment proceeds via an initial seed-and-extend global alignment phase followed by a local search refinement.
• Robustness and Discriminative Power: The spectral signature provides high discriminative power between network structures and yields alignments robust to experimental noise.
• Performance Excellence: Benchmarks report that GHOST recovers larger and more biologically significant shared subnetworks across species compared with contemporary approaches.

Scientific Applications:

• Functional Annotation of Unannotated Proteins: Cross-species network mappings support transfer of functional inference to proteins lacking annotations.
• Discovery of Unobserved Interactions: Conserved network patterns identified by GHOST can suggest potential protein interactions not yet observed experimentally.
• Evolutionary Studies: GHOST enables analysis of evolutionary divergence and convergence of protein complexes and pathways across species.

Methodology:

GHOST computes spectral signatures to assess topological similarity, performs a seed-and-extend global alignment, and refines alignments using a local search procedure.