A Genomic Perspective on Polycystic Kidney Diagnosis: Insights from Gene Expression Analysis

Kidney cysts are fluid-filled sacs that form in the kidneys and may indicate underlying conditions such as polycystic kidney disease (PKD). Understanding the molecular basis of kidney cyst formation is critical, as these cysts can impair kidney function and lead to chronic kidney disease (CKD) or renal failure. This study adopts a genomic perspective to explore polycystic kidney diagnosis by leveraging machine learning techniques to analyze gene expression data related to kidney cyst conditions. Specifically, the paper pursuits a Multi-Layer Perceptron (MLP) model to identify gene expression patterns associated with PKD. A crucial aspect of this research is data preparation, where raw.CEL files from publicly available gene expression datasets are processed using R-based normalization techniques to remove background noise and enhance data quality. Variance-based feature selection is applied to retain biologically significant genes, ensuring a robust input for analysis. The study incorporates dimensionality reduction techniques such as Principal Component Analysis (PCA) to visualize gene expression variability and identify key differentiators between cyst-forming and non-cyst-forming samples. The MLP classifier, trained with increasing epochs, demonstrated a peak accuracy of $86.61 \%$, highlighting the potential of computational methods to support genomic-level understanding in PKD diagnosis.