Unraveling miRNA-Driven DNA Damage Response Networks in Pancreatic Adenocarcinoma: A Multi-Omics and Machine Learning Approach

Abstract

Due to the late detection, aggressive nature, and paucity of treatment options, pancreatic adenocarcinoma (PAAD) remains one of the most lethal cancers worldwide. MicroRNAs (miRNAs) are small non-coding RNAs that function as oncogenes or tumor suppressors. These are highly stable in the blood circulation, hence increasingly recognized as promising biomarkers for early cancer detection. We hypothesize that certain "driver miRNAs" regulate the expression of DNA Damage Response (DDR) genes. To test this, we have built a machine learning model to integrate multi-omics data from The Cancer Genome Atlas (TCGA) PAAD cohort, including copy number alterations, DNA methylation, gene expression, and microRNA (miRNA) expression, along with curated transcription factor interactions. We predicted around 2,000 miRNA-DDR gene interaction pairs; several of these were significantly enriched in a previously known experimentally validated miRNA-target interactions. Our analysis revealed several oncogenic and tumor-suppressive miRNAs that closely correlated with patient outcomes. Importantly, we were able to identify DDR genes that were highly targeted by multiple miRNAs and strongly correlated with patient overall survival. These findings enhance our understanding of the molecular mechanisms in PAAD and open new avenues for using miRNAs as disease biomarkers and suggest target genes for precision oncology.