Automated inference procedure for the determination of cell growth parameters

Abstract

The growth rate and carrying capacity of a cell population are key to the characterization of the population's viability and to the quantification of its responses to perturbations such as drug treatments. Accurate estimation of these parameters necessitates careful analysis. Here, we present a rigorous mathematical approach for the robust analysis of cell count data, in which all the experimental stages of the cell counting process are investigated in detail with the machinery of Bayesian probability theory. We advance a flexible theoretical framework that permits accurate estimates of the growth parameters of cell populations and of the logical correlations between them. Moreover, our approach naturally produces an objective metric of avoidable experimental error, which may be tracked over time in a laboratory to detect instrumentation failures or lapses in protocol. We apply our method to the analysis of cell count data in the context of a logistic growth model by means of a user-friendly computer program that automates this analysis, and present some samples of its output. Finally, we note that a traditional least squares fit can provide misleading estimates of parameter values, because it ignores available information with regard to the way in which the data have actually been collected.

A Multi-center Study on the Reproducibility of Drug-Response Assays in Mammalian Cell Lines

“…It is known that CellTiter-Glo and direct cell counts are poorly correlated when drugs cause large changes in cell size or alter ATP metabolism, thereby changing the relationship between ATP level in a cell extract and viable cell number ( Figure 2B for Palbociclib) (Harris et al, 2016a;Salani et al, 2013;Soliman et al, 2016). The magnitude of this effect depends on the drug being assayed and also on the cell line (Niepel et al, 2017); as a consequence, direct cell counting and CellTiter-Glo can be substituted for each other in some cases but not in others.…”

A multi-center study on factors influencing the reproducibility ofin vitrodrug-response studies

“…It is known that CellTiter-Glo® and direct cell counts are poorly correlated when drugs cause dramatic changes in cell size or alter ATP metabolism, thereby changing the relationship between ATP level in a cell extract and viable cell number ( Figure 2B for Palbociclib) (Harris et al, 2016a;Salani et al, 2013;Soliman et al, 2016). The magnitude of this effect depends on the drug being assayed and also on the cell line ; as a consequence, direct cell counting and CellTiter-Glo® can be substituted for each other in some cases but not in others.…”

Inferring time-dependent population growth rates in cell cultures undergoing adaptation

“…There are other approaches towards finding these growth parameters based on various methods, e.g. steepest descent optimization [11], and different means of Bayesian inference [12,13]. Some models use highly detailed representations of the cell cycle [14].…”