Integrative Analysis of Genome Data Using Deep Embedded Clustering to Identify Population Stratification and Functional Gene Modules

AbstractIntroduction: Next-Generation Sequencing (NGS) data analysis faces computational challenges due to high dimensionality. Traditional clustering methods fail to capture biological complexity in human genetic variation. This study introduces Deep Embedded Clustering (DEC) for genomic pattern discovery.Methods: This study suggests a comprehensive DEC framework applied to 1000 Genomes Project NGS data. Performance was evaluated against conventional clustering algorithms using Adjusted Rand Index (ARI) across varying cluster configurations. Pathway enrichment analysis assessed biological relevance.Results: DEC achieved highly efficient recovery of population structure (ARI=0.892±0.015) with robust performance across cluster numbers. Identified clusters showed significant enrichment for population-specific adaptations: lactase persistence (FDR=1.1×10⁻¹⁰), alcohol metabolism (FDR=2.3×10⁻⁷), and malaria resistance (FDR=3.4×10⁻¹²).Discussion: DEC improves conventional clustering by integrating dimensionality reduction with cluster assignment, revealing biologically meaningful patterns missed by traditional methods. This bridges computational methodology with functional genomics interpretation, though validation in diverse cohorts is warranted.Conclusion: DEC offers an unsupervised learning framework for genomics, enabling biologically meaningful pattern discovery beyond statistical clustering. Our reproducible pipeline provides a foundation for functional module identification in complex genomic datasets.