Corbi

Corbi performs biological network alignment and querying to predict protein–protein interactions and identify conserved subnetworks by integrating biological and structural similarity.

Key Features:

• Network alignment and querying: Extends CNetQ, a network querying method based on conditional random fields (CRF), to address network alignment problems.
• Iterative bi-directional mapping strategy: Employs an iterative bi-directional mapping strategy to improve alignment accuracy between networks.
• Balanced similarity integration: Balances biological and structural similarity measures to avoid structurally- or biologically-dominated alignments.
• CNetA algorithm: Implements CNetA as the alignment algorithm evaluated within the Corbi framework.
• Benchmarking and evaluation: Includes comparisons with four other methods and evaluation using four structural and five biological measures.
• Datasets: Validated on fifty simulated and three real protein–protein interaction (PPI) network alignment instances.
• Implementation: Provided as an R package implementation.

Scientific Applications:

• Protein–protein interaction prediction: Predicts novel PPIs by transferring interaction information across aligned networks.
• Conserved subnetwork identification: Identifies larger conserved subnetworks to reveal functional similarities and evolutionary relationships.

Methodology:

CNetQ uses conditional random fields (CRF) for network querying; CNetA applies an iterative bi-directional mapping strategy; methods were compared against four other algorithms across fifty simulated and three real PPI alignment instances and evaluated with four structural and five biological measures, assessing node and network alignment accuracy.