Detecting differential growth of microbial populations with Gaussian process regression

Abstract

Microbial growth curves are used to study differential effects of media, genetics, and stress on microbial population growth. Consequently, many modeling frameworks exist to capture microbial population growth measurements. However, current models are designed to quantify growth under conditions that produce a specific functional form. Extensions to these models are required to quantify the effects of perturbations, which often exhibit non-standard growth curves. Rather than fix expected functional forms of different experimental perturbations, we developed a general and robust model of microbial population growth curves using Gaussian process (GP) regression. GP regression modeling of high resolution time-series growth data enables accurate quantification of population growth, and can be extended to identify differential growth phenotypes due to genetic background or stress. Additionally, confounding effects due to experimental variation can be controlled explicitly. Our framework substantially outperforms commonly used microbial population growth models, particularly when modeling growth data from environmentally stressed populations. We apply the GP growth model to a collection of growth measurements for seven transcription factor knockout strains of a model archaeal organism, Halobacterium salinarum. Using these models fitted to growth data, two statistical tests were developed to quantify the differential effects of genetic and environmental perturbations on microbial growth. These statistical tests accurately identify known regulators and implicate novel regulators of growth under standard and stress conditions. Furthermore, the fitted GP regression models are interpretable, recapitulating biological knowledge of growth response while providing new insights into the relevant parameters affecting microbial population growth.