Models of the Gene Must Inform Data-Mining Strategies in Genomics

The gene is a fundamental concept of genetics, which emerged with the Mendelian paradigm of heredity at the beginning of the 20th century. However, the concept has since diversified. Somewhat different narratives and models of the gene developed in several sub-disciplines of genetics, that is in classical genetics, population genetics, molecular genetics, genomics, and, recently, also, in systems genetics. Here, I ask how the diversity of the concept impacts data-integration and data-mining strategies for bioinformatics, genomics, statistical genetics, and data science. I also consider theoretical background of the concept of the gene in the ideas of empiricism and experimentalism, as well as reductionist and anti-reductionist narratives on the concept. Finally, a few strategies of analysis from published examples of data-mining projects are discussed. Moreover, the examples are re-interpreted in the light of the theoretical material. I argue that the choice of an optimal level of abstraction for the gene is vital for a successful genome analysis.