Prediction of therapeutic targets and drug candidates for pancreatic ductal adenocarcinoma by network-based computational framework

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies, with limited therapeutic options and an urgent need for novel treatment strategies. This study presented a network-based computational framework for in silico genome-wide therapeutic target identification and drug discovery. A network-based computational framework was developed that integrates transcriptomic data from multiple independent PDAC cohorts to construct gene co-expression networks. Prognostic gene signatures were incorporated to identify feature modules via hypergeometric enrichment analysis. Key regulatory genes within these modules were prioritized using multi-metric centrality analysis. Functional relevance of candidate targets was assessed using single-cell ribonucleic acid (RNA) sequencing, genome-wide clustered regularly interspaced short palindromic repeats (CRISPR) dependency datasets, and clinical survival analyses. To enable therapeutic translation, this study implemented a transcriptome-guided drug discovery strategy to systematically match pancreatic ductal adenocarcinoma (PDAC)-specific gene expression signatures with compound-induced transcriptional profiles from the Library of Integrated Network-based Cellular Signatures (LINCS) database. Candidate compounds were ranked using a cross-cohort consensus scoring approach and subsequently evaluated in vitro using quantitative reverse transcription polymerase chain reaction (qRT-PCR), Western blotting, and cell viability assays. Comprehensive analysis of multi-cohort PDAC transcriptomic data identified 14 key genes with high network centrality, including three potential oncogenic drivers: Cytoskeleton associated protein 2 like ( CKAP2L ), kinesin family member 18A ( KIF18A ), and minichromosome maintenance 10 replication initiation factor ( MCM10 ), which exhibited tumor-specific overexpression and poor prognostic associations. Single-cell RNA sequencing (scRNA-seq) revealed their predominant expression in epithelial tumor cells, while CRISPR screening confirmed functional essentiality. The transcriptome-guided drug discovery approach matched these targets with LINCS perturbation profiles, identifying mitoxantrone/OTSSP167 ( CKAP2L inhibitors), PF-3758309 ( KIF18A inhibitors), and vorinostat ( MCM10 inhibitors) as candidate therapeutics. Experimental validation using qRT-PCR, Western blotting, and cell counting kit-8 (CCK-8) assays collectively demonstrated the dose-dependent therapeutic effects of these compounds, showing significant mRNA suppression, corresponding decreases in protein levels, and potent anti-proliferative activity in PDAC cells. This study establishes a robust, network-based framework for therapeutic target discovery and transcriptome-driven drug prediction in PDAC. It identifies three novel, functionally essential oncogenic drivers and proposes clinically actionable compounds, offering a scalable approach for addressing undruggable targets in cancer therapy.