iPiDA-GCN

iPiDA-GCN predicts associations between piwi-interacting RNAs (piRNAs) and diseases using graph convolutional networks applied to piRNA sequence data, disease semantic data, and known piRNA-disease associations.

Key Features:

• Graph Convolutional Networks (GCNs): Uses GCNs to capture complex and nonlinear relationships between piRNAs and diseases and to mitigate limited training data and insufficient association representation.
• Data Integration: Constructs graphs from piRNA sequence data, disease semantic data, and known piRNA-disease associations.
• Asso-GCN and Sim-GCN: Employs two specialized GCNs—Asso-GCN to extract association patterns from the piRNA-disease interaction network and Sim-GCN to learn from similarity networks.
• Output Module: Applies full connection networks and inner product mechanisms to generate predictive scores for potential piRNA-disease associations.

Scientific Applications:

• piRNA-disease association discovery: Identifies novel piRNA-disease associations to expand known interaction networks.
• Disease mechanism analysis: Supports elucidation of molecular underpinnings of diseases at the RNA level.
• Therapeutic target research: Facilitates development of targeted therapeutic strategies by revealing piRNA involvement in disease.

Methodology:

Constructs graphs from piRNA sequences, disease semantic data, and known associations; applies Asso-GCN and Sim-GCN to extract features; and uses full connection networks with inner-product operations to compute association prediction scores.