2012/10/17 by Chris Lee, Lee, Christopher J., Marc Harper +1
Biochemistry, Genetics and Molecular Biology · Computer Science · #94A15 #Evolution and Genetic Dynamics #Evolutionary Algorithms and Applications #FOS: Computer and information sciences #Gene Regulatory Network Analysis #Information Theory (cs.IT)
paper · pdf · doi:10.48550/arxiv.1210.4808
openalex publication_date 2012/10/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In this paper we outline some mathematical questions that emerge from trying\nto "turn the scientific method into math". Specifically, we consider the\nproblem of experiment planning (choosing the best experiment to do next) in\nexplicit probabilistic and information theoretic terms. We formulate this as an\ninformation measurement problem; that is, we seek a rigorous definition of an\ninformation metric to measure the likely information yield of an experiment,\nsuch that maximizing the information metric will indeed reliably choose the\nbest experiment to perform. We present the surprising result that defining the\nmetric purely in terms of prediction power on observable variables yields a\nmetric that can converge to the classical mutual information measuring how\ninformative the experimental observation is about an underlying hidden\nvariable. We show how the expectation potential information metric can compute\nthe "information rate" of an experiment as well its total possible yield, and\nthe information value of experimental controls. To illustrate the utility of\nthese concepts for guiding fundamental scientific inquiry, we present an\nextensive case study (RoboMendel) applying these metrics to propose sequences\nof experiments for discovering the basic principles of genetics.\n