Severe testing with high-dimensional omics data for enhancing biomedical scientific discovery

Abstract

High-throughput omics experiments provide a wealth of data for exploring biomedical questions and for advancing translational research. However, despite this great potential, results that enter the clinical practice are scarce even twenty years after the completion of the human genome project. For this reason in this paper, we revisit problems with scientific discovery commonly summarized under the term reproducibility crisis. We will argue that the major problem that hampers progress in translational research is threefold. First, in order to establish biological foundations of disorders or general complex phenotypes, one needs to embrace emergence. Second, there seems to be confusion about the underlying hypotheses tested by omics studies. Third, most contemporary omics studies are designed to perform what can be seen as incremental corroborations of a hypothesis. In order to improve upon these shortcomings, we define a severe testing framework (STF) that can be applied to a large number of omics studies for enhancing scientific discovery in the biomedical sciences. Briefly, STF provides systematic means to trim wild-grown omics studies in a constructive way.