Cross-Study Replicability in Cluster Analysis

In cancer research, clustering techniques are widely used for exploratory analyses and dimensionality reduction, playing a critical role in the identification of novel subtypes, often with direct implications patient management. As data collected by multiple research groups grows, it is increasingly feasible to investigate replicability procedures, that is, their ability consistently recover biologically meaningful clusters across several datasets. this paper, we review existing methods assess analyses, discuss framework evaluating cross-study replicability, useful when two or more studies available. These approaches can be applied any algorithm employ different measures similarity between partitions quantify globally (i.e. whole sample) as well locally individual clusters). Using experiments on synthetic real gene expression data, illustrate utility metrics evaluate if same identified collection