Multiresolution Clustering of Genomic Data

Cluster analysis is a widely used unsupervised learning technique in genomic data analysis, with critical applications such as inferring genetic population structures and annotating cell types from single-cell RNA-seq data. However, most existing clustering methods focus on identifying a single optimal partition while overlooking intrinsic relationships among the inferred clusters. Moreover, clustering results produced by different algorithms often appear inconsistent, and there is a lack of principled approaches to extract shared, biologically meaningful patterns across diverse clustering outputs. In this work, we introduce a computational framework that enables systematic exploration of multi-resolution clustering structures in genomic data, starting from an initial configuration generated by any available clustering algorithm. The proposed algorithm provides a unified and principled approach for uncovering complex, nested biological relationships and reconciling discrepancies among clustering results. We demonstrate the utility of our framework through comprehensive simulations and applications to both genetic and single-cell transcriptomic datasets, highlighting its ability to recover interpretable and reproducible clustering structures. Furthermore, we show that our multi-resolution cluster analysis of complex genomic data yields valuable insights into patterns of human population migration and cell differentiation trajectories.