Using fluorescence flow cytometry data for single-cell gene expression analysis in bacteria

Abstract

Fluorescence flow cytometry is a highly attractive technology for quantifying single-cell expression distributions in bacteria in high-throughput. However, so far there has been no systematic investigation of the best practices for quantitative analysis of such data, what systematic biases exist, and what accuracy and sensitivity can be obtained. We here investigate these issues by systematically comparing flow cytometry measurements of fluorescent reporters in E. coli with measurements of the same strains in microscopic setups and develop a method for rigorous quantitative analysis of fluorescence flow cytometry data.We find that forward and side scatter cannot be used to reliably estimate cell size in bacteria. Second, we show that cytometry measurements contain a large shot noise component that can be easily mistaken for intrinsic noise in gene expression, and show how calibration measurements can be used to correct for this measurement shot noise.To aid other researchers with quantitative analysis of flow cytometry expression data in bacteria, we distribute E-Flow, an open-source R package that implements our methods for filtering cells based on forward and side scatter, and for estimating true biological expression means and variances from the fluorescence signal. The package is available at https://github.com/vanNimwegenLab/E-Flow.

Noise propagation shapes condition-dependent gene expression noise inEscherichia coli

“…To estimate mean and variance we used a method that uses forward and side scatter to identify viable cells and fits the logfluorescence distribution by a mixture of a Gaussian and uniform distribution to remove possible outliers (e.g. contaminants, non-growing cells) Galbusera et al , 2019. Replicate measurements performed on different days were highly reproducible, with Pearson correlations R 2 > 0.99 for the mean between replicates in all conditions, and correlations for the variance ranging from R 2 = 0.85 to R 2 = 0.95 (Suppl. Fig.…”

“…The Poissonian term, whose magnitude we denote by b c and is often referred to as the 'intrinsic noise' term, could in principle derive from intrinsic expression noise whose magnitude scales proportional to mean expression Taniguchi et al , 2010;Sánchez and Kondev, 2008. However, by comparing microscopy and flow cytometry measurements we have recently shown that, at these expression levels, the component b c derives almost entirely from the measurement noise of the flow cytometer Galbusera et al , 2019. As shown in Suppl.…”

“…For each strain we collected 5 × 10 4 events. We used a Bayesian procedure that removes outliers to extract the mean and variance of the log-fluorescence distributions as described in Galbusera et al , 2019. Briefly, we first fitted the 4-dimensional signal distribution of forward and side scatter heights and widths by a mixture of a multivariate Gaussian and a uniform 'background' distribution.…”

Corrected score methods for estimating Bayesian networks with error‐prone nodes

“…[9][10][11][12][13] However, measurement errors in flow cytometry data inevitably arise from imperfect measurements, photon-counting statistics, and data storage methods. [14][15][16][17] Figure 1, borrowing from figure 1 in Galbusera et al, 17 illustrates data (such as the height, area, and width of a pulse) reported by a cytometer. To collect such data, one streams cells past a laser light source, and detects them via fluorescence picked up by detectors.…”

“…This technology produces abundant data that can be used to infer cellular signaling networks 9‐13 . However, measurement errors in flow cytometry data inevitably arise from imperfect measurements, photon‐counting statistics, and data storage methods 14‐17 . Figure 1, borrowing from figure 1 in Galbusera et al, 17 illustrates data (such as the height, area, and width of a pulse) reported by a cytometer.…”