WBNPMD

WBNPMD predicts miRNA-disease associations by applying a weighted bipartite network projection that integrates miRNA and disease similarity information to prioritize candidate miRNAs for disease association studies.

Key Features:

• Weighted bipartite network projection: Projects associations between miRNAs and diseases using a weighted bipartite network representation.
• Transfer weights: Constructs transfer weights that integrate known similarities of miRNAs and diseases to configure initial information.
• Two-step bipartite network algorithm: Utilizes a two-step bipartite network algorithm to infer and project miRNA-disease associations.
• Validation methods: Evaluated using leave-one-out cross-validation (LOOCV) and fivefold cross-validation on known miRNA-disease association data.
• Performance metric: Achieved an Area Under the Curve (AUC) of 0.9321 in LOOCV and showed superior predictive accuracy compared to four other state-of-the-art methods.
• Case study predictions: Generated top-50 predicted miRNAs for prostate neoplasm, colorectal neoplasm, and lung neoplasm, with predictions cross-referenced against dbDeMC, miR2Disease, and HMDD V3.0.

Scientific Applications:

• miRNA–disease association prediction: Prioritizes potential miRNA-disease associations for downstream biological validation.
• Disease-specific candidate prioritization: Ranks candidate miRNAs for prostate neoplasm, colorectal neoplasm, and lung neoplasm based on projected association scores.
• Cross-database validation: Facilitates cross-referencing of predicted miRNAs with dbDeMC, miR2Disease, and HMDD V3.0 for corroboration of associations.

Methodology:

Constructs transfer weights integrating known miRNA and disease similarities and applies a weighted, two-step bipartite network projection algorithm to infer miRNA-disease associations.