Dr AFC

Dr AFC predicts drug repositioning candidates by integrating anti-fibrosis characteristics into machine learning models to prioritize compounds for treating fibrotic diseases.

Key Features:

• Anti-Fibrosis Knowledge Base: A curated knowledge base assembled from diverse data sources encoding anti-fibrosis characteristics that underpin predictive modeling.
• Machine Learning Models: Two models—Structural Profile Prediction Model (SPPM) and Biological Profile Prediction Model (BPPM)—trained on structural and biological profiles derived from the knowledge base.
• Validation and Performance: Models were validated on three external public datasets with SPPM AUCs of 0.879 (training) and 0.814 (testing) and BPPM AUCs of 0.972 (training) and 0.874 (testing).
• Exploration of Multi-Targeting Natural Products: Identification and prioritization of multi-targeting natural products that exhibit significant anti-fibrosis characteristics as potential repositioning candidates.

Scientific Applications:

• Drug Repositioning Discovery: Identification of existing drugs with potential efficacy against fibrotic diseases by targeting anti-fibrosis pathways.
• Candidate Prioritization: Prioritization of compounds, including multi-targeting natural products, based on predicted structural and biological anti-fibrosis profiles.
• Cross-Disease Therapeutic Strategies: Support for identifying interventions applicable across multiple fibrotic conditions via shared anti-fibrosis mechanisms.

Methodology:

Construction of an anti-fibrosis knowledge base; derivation of structural and biological profiles; training of SPPM and BPPM machine learning models; validation using three external public datasets with reported AUC metrics.