Computational Identification of Differentially Expressed Genes in Prostate Cancer Using Integrated Public Datasets

Prostate cancer (PC) is a major global health challenge, yet the molecular drivers of its aggressive forms remain incompletely understood. Computational integration of publicly available datasets can identify molecular signatures and candidate biomarkers requiring experimental validation. In this study, I demonstrate that robust transcriptomic analyses can be performed by integrating publicly available RNA-seq datasets, even when control and tumour samples originate from different studies and technical batches. I analysed RNA-seq data from RWPE-1 normal prostate epithelial cells and PC3 androgen-independent prostate adenocarcinoma cells using nf-core pipelines and open-source tools, applying rigorous quality control and correcting for batch effects with Surrogate Variable Analysis (SVA). My differential expression analysis identified 12 337 significantly altered genes (p-adj < 0.05), with 6462 upregulated and 5875 downregulated in PC3 cells. This analysis confirmed established oncogenes such as YBX2 and CX3CL1, and revealed novel candidates in prostate cancer, including the upregulation of PRSS21 and the downregulation of PAX6, THSD7A, and the metabolic regulator PDK4. Functional enrichment analyses consistently highlighted pathways critical to cancer progression, including extracellular matrix organization, cell adhesion, and neural signalling. This work provides a resource of potential biomarkers and therapeutic candidates and demonstrates a methodology that enables researchers to extract possible clinically relevant molecular mechanisms from public data using reproducible workflows.