Ranking Metrics for Genomic Regions

Mining the genomic datasets to extract regions of interest is a difficult task and besides being memory and compute intensive, involves a large amount of domain expertise. Efficient computational approaches to derive insights from these datasets are imperative to their better understanding. In this work, we present an approach to combine information theoretic metrics to rank the genomic regions from two distinct datasets, and ultimately show that these metrics, without insights into the underlying data, not only capture genomic regions of interest but also produce a novel way of comparison among genomic regions across scales. This in turn helps in identifying the scale at which a given region maximizes the information contained in it.