Abstract
Idiopathic pulmonary fibrosis (IPF) is a progressive and fatal lung disease characterized by marked cellular heterogeneity and dysregulated signaling networks, which pose substantial challenges for therapeutic development. To address these complexities, we developed an integrative computational framework that combines machine learning-based ligand prediction, single-cell transcriptomic analysis, genetic causal inference, and structure-based molecular modeling to prioritize multitarget therapeutic candidates for IPF. Using curated molecular activity data sets, target-specific machine learning models were constructed to predict compounds with potential activity against 13 IPF-associated receptor tyrosine kinases implicated in fibrotic remodeling. Several clinically used kinase inhibitors were identified among the top-ranked candidates, suggesting potential multitarget activity profiles. Single-cell RNA sequencing analysis further revealed that these targets are enriched in stromal and immune cell populations involved in fibrotic progression, supporting their biological relevance in the IPF microenvironment. Mendelian randomization and colocalization analyses highlighted PDGFRB as a gene with strong statistical evidence of causal association with IPF susceptibility. Molecular docking and molecular dynamics simulations revealed stable predicted interactions between selected ligands and PDGFRB, providing structural support for the plausibility of these compound-target relationships. Collectively, this integrative framework links computational prediction with transcriptomic and genetic evidence, offering a systematic strategy for prioritizing candidate therapeutics capable of modulating multiple disease-associated pathways in IPF.