Ignoring, Accounting For, and Embracing Noise in Biology
Rochman, Nash Delta
MetadataShow full item record
The universe is a noisy place in every sense of the word. From the detection of cosmic microwave background radiation to the formulation and application of the Heisenberg uncertainty principle, most corners of modern science have some stochastic flavor. Biology, and cell biology in particular, is no exception. Cells have a very large number of ever-changing, moving parts and seem to only rarely lend themselves to simple, deterministic characterization. Having accepted this, how do we move forward - how do we treat this stochasticity when we seek to model and ultimately understand the governing functions of a cell? I would like to discuss three strategies to answer this question. Firstly, we may ignore the noise. While it is always important to first acknowledge variations from mean behavior, sometimes grappling with the prediction of higher moments just distracts from the concept we wish to express. Second, we may “treat” the noise. Often, we are not writing down an elegant theory but pipetting and we just want to know if our results are statistically significant. Slightly better, we may wish to account for some bias in the calculation of a mean value which is unmeasurable but well approximated, like the age distribution of a population. Third, and the most interesting case, we may embrace the noise. Occasionally, the heterogeneity of a population is more revealing of underlying mechanism than even mean behavior, as is especially true when studying adaptability. I will cover, in chronological order, five disparate applications of these three strategies; 1) a model of surface interactions of helical filaments - with a focused example of amyloid beta fibrils - in which noise is ignored; 2) a study of bacterial cell cycle duration regulation - in which noise is embraced and heterogeneity is predicted to confer adaptability to environmental stressors; 3) an investigation into mammalian cell volume regulation - in which noise is treated to gauge statistical significance across ensembles of single cells and age-related heterogeneity is invoked to explain higher moments of observed volume distributions; 4) a study of stem cell pattern formation - in which noise is treated to measure the predictive value of a data-driven model of cell specification; and 5) an investigation into the impacts of ergodicity on cell cycle duration and population growth rate - in which noise is embraced and heterogeneity is predicted to imply asymmetry in division under some commonly assumed conditions.