We develop a symbolic regression framework for extracting the governing\nmathematical expressions from observed data. The evolutionary approach, faiGP,\nis designed to leverage the properties of a function algebra that have been\nencoded into a grammar, providing a theoretical guarantee of universal\napproximation and a way to minimize bloat. In this framework, the choice of\noperators of the grammar may be informed by a physical theory or symmetry\nconsiderations. Since there is currently no theory that can derive the\n'constants of nature', an empirical investigation on extracting these\ncoefficients from an evolutionary process is of methodological interest. We\nquantify the impact of different types of regularizers, including a diversity\nmetric adapted from studies of the transcriptome and a complexity measure, on\nthe performance of the framework. Our implementation, which leverages neural\nnetworks and a genetic programmer, generates non-trivial symbolically\nequivalent expressions ("Ramanujan expressions") or approximations with\npotentially interesting numerical applications. To illustrate the framework, a\nmodel of ligand-receptor binding kinetics, including an account of gene\nregulation by transcription factors, and a model of the regulatory range of the\ncistrome from omics data are presented. This study has important implications\non the development of data-driven methodologies for the discovery of governing\nequations in experimental data derived from new sensing systems and\nhigh-throughput screening technologies.\n