Making our plant modelling community more than the sum of its parts: a personal perspective

Abstract

The rise of mathematical modelling represents a transition in any scientific area towards quantitative and unequivocal presentation of hypotheses and theory. In Physics, phenomena are now predicted from mathematical models and computer simulations of those models well before observations are made to confirm these predictions. Indeed the largest endeavours in Physics, such as particle physics accelerators and astrophysics observational platforms, are created to test the most profound predictions of such models. Given the huge complexity of living organisms coupled with massive species and even phenotype within genotype diversity, we are far from the same level of advancement, yet need to approach it. Complexity of our systems means that many biological modelling efforts will remain, largely, based on emergent properties and phenomena. Nevertheless, complete models of the full complexity of single-celled organisms are beginning to transition to eukaryotes (Beard et al. 2012; Service 2016). Within narrow areas of gene function, we are already seeing successful projections from gene expression to prediction of growth and development of whole plants (Chew et al. 2014). Modelling provides a framework in which we can precisely organize and test our quantitative knowledge and hypotheses about how a plant process or combination of processes works and then test these against reality. As such, it provides a data-hypothesis-test-learn cycle to improve our understanding of plants and their use. Equally, the rapid growth of high-throughput ‘omics facilities is delivering ever-increasing amounts of data for which our capacity and ability to analyse and interpret lags. Mathematical models coupled with high-performance computing provide a means to deliver this needed acceleration. Simultaneously it should provide the means to predict which data is needed most, so providing feedback and focus for ‘omics approaches. This wealth of data also provides unprecedented opportunities for improving the precision of models by high-speed data to model linkage. Similarly, numerical and text mining knowledge discovery offer much to improving mathematical modelling of plant processes, with opportunities for automated improvement of representation and parameterization (Fer et al. 2018). In parallel, computer simulation of mathematical models has evolved from printouts of numbers to 3D representations of the growth and development of organs, whole plants and even communities of plants that can be indistinguishable from the real thing (Fig. 1). This facilitates identification of emergent phenomena while providing unprecedented opportunities in revolutionizing plant science education (Prusinkiewicz et al. 2007; Prusinkiewicz and Runions 2012; Runions et al. 2017).