A Conceptual Approach to Using Relevant Patterns in Genomic Data Analysis

Several models have been proposed to represent human genomic information. An interesting approach for supporting genomic applications for health consists of a two-layer representation. In this approach, high-level concepts describing distinct aspects of the human genome at an abstract level are mapped to data representing actual physical measurements. This two-layer method allows users to formulate high-level queries on the concepts and map them onto real datasets. Additionally, the approach is extensible, allowing new conceptual views corresponding to specific genomic features to be mapped to the lower data layer without impacting previous mappings. We here present how concept-layer and data-layer instances can be composed into patterns corresponding to classic genomic studies: diseases with case-control comparisons, multi-omic representations for the same patients, and comparisons within families for rare genetic diseases. We show that these patterns effectively support genomic data users (i.e., clinicians, geneticists, and bioinformaticians) in genomic analysis practices.