Following Thomas Kuhn's seminal 1962 book that introduced the notion of scientific paradigms, we describe a computational methodology that leverages this concept in a concrete formulation. We report this approach partially as a methodology for framing and scoping the knowledge representation and analysis work necessary to build tools to serve a specific community. However, it also has technical implications for semantic web representations, text mining tools and reasoning about scientific workflows. We explore this viewpoint in two contexts - biomarker studies of neurodegenerative diseases and studies of vaccines - with the strategic intent of developing practical biomedical informatics systems that make experimentally testable predictions. We will discuss the implications of this formalism in developing a structured representation of phenotype that goes beyond a 'hierarchy of physiological phenomena' and its possible application to molecular biology and genetic approaches.