MINIREVIEW

Applied and Environmental Science

Computational Analysis of Microbial Flow Cytometry Data

Peter Rubbens,a Ruben Propsb

Flanders Marine Institute (VLIZ), Ostend, Belgium

a

b Center for Microbial Ecology & Technology (CMET), Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium

ABSTRACT Flow cytometry is an important technology for the study of microbial

communities. It grants the ability to rapidly generate phenotypic single-cell data that
are both quantitative, multivariate and of high temporal resolution. The complexity
and amount of data necessitate an objective and streamlined data processing work-
flow that extends beyond commercial instrument software. No full overview of the
necessary steps regarding the computational analysis of microbial flow cytometry data
currently exists. In this review, we provide an overview of the full data analysis pipe-
line, ranging from measurement to data interpretation, tailored toward studies in mi-
crobial ecology. At every step, we highlight computational methods that are poten-
tially useful, for which we provide a short nontechnical description. We place this
overview in the context of a number of open challenges to the field and offer further
motivation for the use of standardized flow cytometry in microbial ecology research.

KEYWORDS bioinformatics, cytometry, fingerprinting, data analysis, microbial ecology,

single cell, multivariate statistics

F low cytometry (FCM) is a single-cell technology that provides an optical description

of individual particles based on scatter and fluorescence information. Microbial
FCM has a long history, and its first applications in the field date back to the late 1970s

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

to investigate the physiological properties of individual cultures (1, 2). The most preva-
lent application in microbiology remains the quantification of cell population densities
in a wide range of matrices, ranging from lab cultures to marine, freshwater, soil, and
fecal samples (3–8). FCM has been applied to many types of microorganisms, mostly
phytoplankton and bacteria, but other types of microorganisms include single- and
multicellular fungi (9, 10) and viruses (11). For many groups of microorganisms, it has
proven to be both accurate and reproducible and can generate results faster than
Citation Rubbens P, Props R. 2021.
existing plate count and marker gene approaches, such as 16S rRNA gene amplicon Computational analysis of microbial flow
sequencing (6, 12). The development of online and real-time FCM facilitates the quanti- cytometry data. mSystems 6:e00895-20.
fication of microbial community dynamics at a very high temporal resolution (13–15). https://doi.org/10.1128/mSystems.00895-20.

A large body of research exists on extracting biological information, in addition to Editor Pieter C. Dorrestein, University of
California, San Diego
cell enumeration measurements, from the multivariate single-cell data acquired by
The review history of this article can be read
FCM. Phenotypic properties, such as size, shape, morphology, activity, membrane per- here.
meability, pigmentation, and nucleic acid content are measured in various degrees, Copyright © 2021 Rubbens and Props. This is
depending on the applied cell-labeling technique (16, 17). The major ongoing wet-lab an open-access article distributed under the
terms of the Creative Commons Attribution 4.0
FCM developments for microbiology research can be broadly classified into (i) develop- International license.
ment and standardization of novel staining methods (17, 18) and (ii) novel laboratory Address correspondence to Peter Rubbens,
protocols to efficiently extract cells from complex matrices (4, 8). Much less attention is peter.rubbens@vliz.be, or Ruben Props,
ruben.props@ugent.be.
given to computational methods that can assist in the analysis of microbial cytometry
data. As a result, many microbiologists perform manual interventions during their data Computational Analysis of Microbial Flow
Cytometry Data: a step-by-step minireview
analysis, including decisions with respect to denoising, quality control, cell population highlighting the full data analysis pipeline of
identification and statistical analyses, on a sample-by-sample or batch-by-batch basis. This microbial flow cytometry data.
inevitably results in user biases, such as reduced reproducibility, but it can also obscure Published 19 January 2021
meaningful biological information not apparent from the user’s own interpretation. Many

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 1

Minireview

FIG 1 Schematic overview of a flow cytometry analysis. Suspended particles are aligned one by one by hydrodynamic focusing.
Next, each particle is interrogated by one or more lasers in the flow cell. The resulting scatter (FSC and SSC) and fluorescence
signals (denoted “FL”) of each cell are captured by multiple detectors. Fluorescence is measured at multiple wavelength intervals
(three in this illustration). The electronic signals originating from these detectors are then finally transformed into digital ones.

computational methods have emerged in the biomedical research field over the past few
years to address these shortcomings, grouped together under the names “FCM bioinfor-
matics” or “computational FCM” (19, 20). These aim to facilitate and improve the objectiv-
ity, speed, and reproducibility of the data analysis. Likewise, microbiologists have the possi-
bility to set up a dedicated data analysis pipeline to benefit from the same advantages as

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

immunologists do.
In this review, we aim to provide a streamlined overview of the data analysis possi-
bilities along a typical computational workflow for microbial FCM data. A demonstra-
tion of such a workflow in R can be found online at https://rprops.github.io/MSys
_FCMreview/Demo.html. We highlight a number of interesting applications in which
FCM is used to perform ecological studies. Additionally, we try to point to a number of
challenges in the field of microbial FCM that motivate the use and development of
standardized FCM for microbiology research.

MICROBIAL FLOW CYTOMETRY

A basic overview of a flow cytometry analysis is given in Fig. 1. Suspended particles
are first aligned on a one-by-one basis by means of hydrodynamic focusing. Each parti-
cle is then interrogated by one or more lasers. Optical filters allow one to measure
emitted fluorescence at multiple wavelengths, next to forward scatter (FSC) and side
scatter (SSC) signals. Photomultiplier tubes are used to convert the fluorescence and
scatter signals to an electronic signal. The morphology of the cell is reflected in the
FSC (size and shape) and SSC (intracellular complexity). The measured fluorescence is
the result of autofluorescent properties (such as pigments) or the interaction with a flu-
orescent dye. Mostly, generic stains that target properties related to nucleic acid con-
tent, membrane integrity and other physiological aspects, such as lipid content,
enzyme activity, and translational activity, are used (3, 17). The technology is fast in the
sense that it is able to measure more than thousands of particles per second. It is quan-
titative, because each particle is described by a numeric multivariate measurement
that represents a unique optical signature for each particle.

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 2

Minireview

The main applications of FCM are the biological and clinical study of mammalian
cells (21), also known as immunophenotyping FCM. By now, immunophenotyping data
routinely represent large antibody panels of up to 28 individual biomarkers, repre-
sented by 28 different fluorescence parameters (22). These applications form the main
drivers of instrument development and research. Microbial FCM has a number of differ-
ent characteristics and challenges compared to immunophenotyping FCM. First, most
prokaryotic cells are much smaller in size and volume than human or mammalian cells.
Therefore, measurements can lie close to the detection limit of an instrument. Second,
while most cells are small, the size range within which microbial cells occur is larger
than for mammalian cells, covering a range between 0.2 and 500 m m. Third, microbial
communities comprise high levels of phenotypic and phylogenetic complexity (e.g.,
1,000s of taxa) and heterogeneity (16). As such, contrasting results concerning the
establishment of multicolor staining panels for microbial communities have been
reported. Single- and double-staining methods are routinely used (23, 24), with the
majority of research relying on one or two general markers with phenotypic (e.g.,
nucleic acids or membrane permeability) (25) or phylogenetic (e.g., see reference 26)
specificity. It appears much more difficult to standardize and broadly apply a triple-
staining protocol, as the efficiency and stability of cell staining protocols are depend-
ent on the bacterial taxa on which they are applied. Although successful approaches
are reported in the literature (27, 28), issues such as fluorescence instability hamper
their widespread use and further development (23). Therefore, microbial FCM data are
characterized by data with fewer dimensions compared to immunophenotyping FCM.

DATA ANALYSIS
A typical FCM data analysis pipeline can be broadly divided into multiple catego-
ries, of which an overview is given in Fig. 2. These include preprocessing of the data,
visualization, cell enumeration of specific populations or the whole community, cyto-
metric fingerprinting, community-level analysis and data format and storage. While
some steps are necessary, others are optional and depend on the research question
and experimental setup. We have summarized and ordered the computational meth-
ods that we discuss in this minireview (see Table 1). Here, we focus on software pack-

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

ages that are publicly available in the R statistical programming language. Note that a
number of packages are also available in other languages, such as Python or Matlab.

DATA FORMAT
FCM data are stored in flow cytometry standard (FCS) format from commercial soft-
ware. The most recent version, FCS 3.1, was introduced in 2010 (29). The area (A), the
height (H), and sometimes also the width (W) of fluorescence and scatter pulses are
recorded for each individual particle. In addition, the file format also allows one to
store metadata describing the experimental settings.

PREPROCESSING
Before FCM data can be analyzed, a number of preprocessing steps need to be per-
formed. Some of them are optional; others are recommended or necessary. The basic
steps are available within the generic FCM software package flowCore (30).
Compensation. If a detector has an optical filter that captures signals coming from
multiple stains, false-positive cells can be detected. Data compensation intends to cor-
rect for emission signal spillover from one stain (e.g., Syto59) into the channel desig-
nated for another stain (e.g., propidium iodide). Currently, this is only rarely applied
due to the limited availability of multicolor FCM protocols to analyze microbial commun-
ities, although a few examples can be found in the literature (23, 31–34). Functions to
perform compensation are incorporated into the flowCore package.
Transformation. Fluorescence and scatter values registered for microbial cells can
differ by orders of magnitude and need to be transformed to enable separation of
instrument noise and cell signals. These values exhibit linear behavior at small scales,

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 3

Minireview

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

FIG 2 Overview of a reproducible microbial FCM data analysis pipeline. First, the data are preprocessed in several steps (compensation, transformation, noise
removal, and quality control). Second, the data are analyzed, which can be done in multiple ways: through visualization, determination of cell concentration,
cell population identification, and/or cytometric fingerprinting. Once processed, the data can be analyzed at the community level to make ecological
inferences. Upon conclusion of the analysis, the raw data are properly annotated and stored in a publicly accessible database. FP, false positives; TN, true
negatives; TP, true positives; QC1 to -4, quality control measurements, respectively; pop, population; PCoA 1 and -2, principal coordinate axis 1 and 2,
respectively.

but as they increase, their values increase exponentially, resulting in values that are
orders of magnitude larger. Traditionally, microbial FCM data are transformed using a
logarithmic function. However, measurements can be negative as well, making the log-
arithm unsuitable to transform the data in this case. More advanced transformations
are recommended, such as the arcsine hyperbolic function or a generalized extension
with one or more adjustable parameters, often referred to as the “biexponential” or
“Logicle” transformation (35, 36). These transformations are available within the
flowCore package. Forward and side scatter information can also be analyzed on a lin-
ear scale for heterogeneous cell populations of larger cells (37), but most microbiologi-
cal applications require a transformation of the scatter parameters as well.
Noise removal. Instrument noise is always present in the data caused by the mea-
surement of (in)organic particles, cell aggregates and electronic noise. Due to the small
cell sizes of microbes, live cells can have fluorescence and scatter properties in a man-
ner similar to the instrument noise. Therefore, noise removal is often performed

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 4

Minireview

TABLE 1 Overview of peer-reviewed computational methods for performing data preprocessing, visualization, cell population identification,
cytometric fingerprinting and data storage
Reference(s) (applied
Category Method (reference) Short description to microbial data?)a
Preprocessing flowCore (30) Basic data manipulation, gating, compensation, and transformation 12, 72–74
Preprocessing flowTrans (37) Optimized individual channel transformations
Preprocessing flowAI (42) Automated denoising 75, 76
Preprocessing flowClean (41) Automated denoising
Preprocessing flowStats (77) Per-channel normalization 78
Visualization flowViz (38) Customized FCM data visualization 9, 79, 80
Visualization ggcyto (39) Customized FCM data visualization with ggplot-like functionality 74
Visualization viSNE (81) Dimensionality reduction and visualization using t-SNE 59, 82
Visualization UMAP (83) Dimensionality reduction using UMAP
Cell population identification flowClust (84) t-distribution mixture model with Box-Cox transformation 85
Cell population identification flowEMMI (54) Gaussian mixture model with the Bayesian information criterion 54
Cell population identification flowPeaks (86) k-means clustering followed by peak search and merging using a 87, 88
Gaussian mixture model
Cell population identification flowDensity (89) Density-based sequential bivariate gating 54, 90
Cell population identification FlowSOM (52) Self-organizing map and meta-clustering
Cell population identification PhenoGraph (53) k-nearest neighbor weighted graph and Louvain method for 82
community detection
Cytometric fingerprinting CHIC (61) Two-channel histogram image comparison 61, 72, 91
Cytometric fingerprinting flowCyBar (60) Manual annotation of interesting regions 60, 91–93
Cytometric fingerprinting flowDiv (62) Fixed-binning grid over multiple two-channel combinations 62
Cytometric fingerprinting flowFP (64) Distribution-dependent binning in hyper-rectangles 91, 94–96
Cytometric fingerprinting Phenoflow (12) Fixed-binning grid and kernel density estimation over multiple two- 12, 76, 97, 98
channel combinations
Cytometric fingerprinting PhenoGMM (67) Overclustering using a Gaussian mixture model 67, 99
Data storage FlowRepository (70) Public database to store and annotate FCM data 34, 75, 100, 101
aThis column highlights references in which the method has been applied to microbial FCM data.

manually, by defining rectangle, quadrant, ellipsoid, or generic polygon regions (i.e.,

“gates”) containing the cell signals. As stated in the introduction, this can be laborious
in time and introduce subjective biases (20). At the individual-sample level, optimal gates
can often differ due to intersample variation. It is advised to use, whenever possible, the
same gating template within a single experiment, although samples analyzed with differ-

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

ent gating templates can still be compared using proper control samples. Negative-con-
trol samples are then necessary, which can include heat-killed samples and 0.2-m m-fil-
tered samples, either stained and/or unstained. A gating template can be defined using
the flowCore package. Denoising is usually guided by user-dependent iterative visualiza-
tion of the scatterplots. FCM data visualization is supported by the flowViz and ggcyto
packages (38, 39). The number of gating steps are largely dependent on the complexity
of the analyzed sample. High degrees of noise may require additional signal filtering from
multiple fluorescence (e.g., autofluorescence on violet laser) and/or scatter channels. In
most measurements, cell aggregates, such as doublets, triplets, or chains, are measured
as well and can be identified through visualizing the area and height parameter of the
primary fluorescence or scatter channel. However, for microbiological applications, it
remains difficult in practice, and there is no consensus yet on how to best handle cell ag-
gregate signals. We recommend the optimization of sample preparation protocols to
reduce the percentage of cell aggregates; these can include the use of filtration, ultrasoni-
cation, surfactants (e.g., Tween, Triton X-100), complexing agents (e.g., EDTA, sodium
pyrophosphate), and/or Ca21/Mg21-free buffers (4, 40).
Quality control. The quality of the data and its acquisition are subject to both
instrument and biological variation. The ideal data acquisition consists of the measure-
ment of homogeneous and stable cell signal distribution during sample analysis.
However, deviations can occur, for example, due to large particles that clog the system
or air bubbles that cause gaps in the data. This results in aberrations in the data, such
as spikes, gaps or gradual degradation of the mean fluorescence intensity. These need
to be addressed and, depending on the research question, removed. While these

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 5

Minireview

actions can be done manually, algorithms to detect and remove anomalies automati-
cally by inspecting individual cell parameters in the function of the acquisition time
have been developed. By applying statistical methods in combination with anomaly
detection strategies, deviating segments are annotated and removed. A number of
methods exist; see, for example, flowClean, flowAI and flowCut (41–43). Further research
is needed to evaluate these algorithms for cell counting, as they can drastically influence
the number of cells measured. As such, we currently do not recommend applying these
quality control algorithms for cell counting, but we do recommend them for fingerprint-
ing and cell population identification applications.

CELL POPULATION IDENTIFICATION

After the data preprocessing, samples can be further analyzed. The most common
analysis in microbial FCM is to characterize the microbial load by enumerating micro-
bial cell densities of the total community, quantified as the number of particles per
milliliter or gram (44).
However, the data can contain distinct cell populations caused by differences in cell
size, morphology, and autofluorescent properties (e.g., phytoplankton [45]) or due to the
use of specific stains (e.g., a nucleic acid stain to detect nucleic acid populations in aquatic
environments [46]). While these are routinely gated manually, cell population identification
algorithms detect these automatically and therefore reduce the bias and analysis time in-
herent in manual gating procedures by experts (20). Dimensionality reduction algorithms
can be used to visualize the multivariate single-cell data at once and to explore whether
distinct cell populations are present in the data. These include principal-component
analysis (PCA), but more advanced algorithms have demonstrated their advantages for
immunophenotyping cytometry, such as t-distributed stochastic neighbor embedding
(t-SNE) and uniform manifold approximation and projection (UMAP) (47, 48).
The performance of cell populations identification algorithms has been thoroughly
benchmarked in terms of cluster accuracy, stability, rare cell type discovery and com-
puting time using standardized immunophenotyping FCM or mass cytometry data sets
(49–51). FlowSOM (52) has been proposed as the least time-intensive algorithm with
favorable results for human mass cytometry data (50), with PhenoGraph (53) being an

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

interesting competitor. Recently, flowEMMI (54), a clustering approach based on
Gaussian mixture models and the expectation-maximization algorithm, has been pro-
posed and compared to a number of additional algorithms to identify clusters in two-
channel bacterial samples. Another option is to perform single-cell classification to iden-
tify known bacterial populations (55–58). These can be helpful in case it is known which
populations are present in the data and one expects their properties to remain stable
throughout the experiment; however, especially the latter is often difficult, due to the
phenotypic heterogeneity of bacterial populations (59).

CYTOMETRIC FINGERPRINTING
The second set of algorithms falls under the category of cytometric fingerprinting
approaches. In this case, the focus lies on modeling the multivariate distribution of sin-
gle-cell observations by dividing the parameter space into regions in which cell counts
or densities are recorded. The identification of distinct cell populations is, in this case, a
secondary objective. Three categories of cytometric fingerprinting approaches can be
distinguished, based on how these regions are determined.
 For manual approaches, multiple clusters or gates are manually drawn in regions
of interest and applied to all samples (see the FlowCyBar algorithm [60]).
 For fixed-binning approaches, a grid of dimensions L by L with equally sized bins is
placed over one or multiple bivariate channel combinations, and the cell count
per bin is registered. Cytometric histogram image comparison (CHIC), Phenoflow,
and flowDiv have been specifically developed for microbial cytometry data (12,
61, 62).

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 6

Minireview

 For adaptive-binning approaches, a grid or other structure with various region

sizes and shapes is placed over a bivariate or multivariate combination of FCM
parameters. The size and shape depend on the distribution of the data, with typi-
cally small bin sizes for those regions of high density and vice versa. The first
adaptive-binning approach, termed probability binning, was already proposed
in 2001 and divides the data in hyper-rectangular bins of various sizes (63). The
algorithm is publicly available as a software package under the name of flowFP
(64). An extension, called PB-sQF, that uses probability binning in combination
with the quadratic-form distance statistic to compare two samples has been
developed (65). Recent alternatives include the search for local density peaks, af-
ter which bins are created using Voronoi tessellation (66), and PhenoGMM, an
approach that uses all multivariate information at once by overclustering the
data using a predefined large number of Gaussian mixtures (67).

Limited research has been devoted to a comparison of fingerprinting methods.

Therefore, it is difficult to provide a clear recommendation on which method(s) a user
should use for their data. In terms of time of analysis and objectivity, operator-independ-
ent methods are preferred over manual methods. Fixed-binning approaches, such as
PhenoFlow and CHIC, model the distribution of the data by using a two-dimensional
gridded approach. However, these approaches become less performant when a user
wants to incorporate more parameters. In addition, the number of community-describ-
ing variables is large. Adaptive-binning approaches require some time to estimate the
gating template. However, these are more advantageous to model multivariate data and
result in fewer community-describing variables.
Some years ago, the FlowCAP (flow cytometry, critical assessment of population
identification methods) initiatives were organized within the immunophenotyping
cytometry community (49, 68). In this initiative, a number of highly curated data sets were
provided to objectively compare cell population identification algorithms. Currently, these
data sets are still the standard to benchmark new computational methods, and their meth-
odology forms the basis for more recent benchmark studies (50, 51). Microbial FCM cur-
rently lacks highly curated data sets. These would be, in combination with a set of com-
monly agreed-upon data analysis objectives, of great value for the development of

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

cytometric fingerprinting methods.

DATA STORAGE
It is recommended that the raw data be stored in FCS format in public FCM reposi-
tories, such as Cytobank or FlowRepository (69, 70). The corresponding accession iden-
tifiers should be added to the data section of every publication and not just the final-
count table. FlowRepository is recommended by multiple journals and societies,
including Cytometry Part A (the official journal of the International Society for the
Advancement of Cytometry [ISAC]), all American Society for Microbiology and PLOS
journals, and Springer Nature. Another helpful tool is the minimum information about
a flow cytometry experiment (MIFlowCyt) document, which assists in the annotation
of the minimum of information that is required to report an FCM experiment (71). A
minor caveat is that current guidelines are tailored toward biomedical experiments.
MIFlowCyt is incorporated in FlowRepository. An overview of peer-reviewed compu-
tational methods for performing data preprocessing, visualization, cell population
identification, cytometric fingerprinting, and data storage can be found in Table 1.

ECOLOGICAL INFERENCES AND APPLICATIONS

The output of cell population identification and fingerprinting algorithms are con-
tingency tables of counts or densities across the determined multivariate regions,
which may be bins, clusters or manually selected regions. With these tables, a variety of
traditional multivariate methods (e.g., PCA, canonical-correspondence analysis [CCA],
permutational multivariate analysis of variance [PERMANOVA], etc.) can be applied to
test for differences among sample groups in the function of experimental conditions

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 7

Minireview

(e.g., pH, nutrient concentrations, host disease state, etc.) (102, 103). Traditional ecological
parameters, such as alpha and beta diversity, and a range of stability and functional diver-
sity metrics have been developed that enable researchers to quantitatively compare
changes in community structure (104, 105). The availability of reference bead data with
refractive indices matching those of bacterial cells and microscopy-based validation
experiments can be used to create predictive models for cell size and biovolumes of indi-
vidual cells and populations (88, 106–108).
For all the described algorithms and metrics, there exist numerous research applica-
tions, most of which are situated in the aquatic research domain, although research in air,
soil, sediment, and clinical microbiology is gaining traction (4, 7, 109, 110). A few studies
have used cell population identification methods to analyze microbial FCM data. These
include the identification of physiological populations (i.e., high and low nucleic acid pop-
ulations in marine and freshwater systems) (85, 111), phytoplankton populations (88, 90),
or different strains of yeast (9). In contrast, fingerprinting methods have been more
broadly applied in environmental microbiology, where they have been used to track
changes in drinking water, sludge, and soil microbiome structures over time and in func-
tions of environmental conditions (15, 92, 94, 112). In a clinical setting, fingerprinting has
been used to infer bactericide treatment effects in saliva microbiomes (93), to train a pre-
dictive model for Crohn’s disease in gut microbiomes (99), and to test for antibiotic sus-
ceptibility (65, 113). FCM has also proven to be a complementary technique to next-gen-
eration sequencing technologies to enable absolute quantification of microbial taxa
(114). Even more, the taxonomic community structure based on 16S rRNA gene amplicon
sequencing has been associated with cytometric community structures in multiple envi-
ronments, including freshwater (12, 115), marine (62, 116), and gut communities (99).

CONCLUSIONS
Microbial FCM applications are rapidly evolving, for example through the use of online
and real-time FCM (13, 14), the development of polychromatic staining panels dedicated
to microbial research (28), and integration with molecular analyses (101, 117, 118). The
amount and complexity of the data will continue to increase as the technology is inte-
grated further into clinical, environmental and industrial research. This will necessitate the
need for objective and streamlined bioinformatics workflows to achieve a quantitative

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

and reproducible data analysis. The collaboration across cytometry disciplines will be cru-
cial to ensure the adoption of computational methods by a wider user base in the field of
microbiology. We hope with this review to have contributed to this end and look forward
to new developments that are yet to emerge in the field.

SUPPORTING INFORMATION
A demonstration of a computational workflow can be found at https://github.com/
rprops/MSys_minireview.

ACKNOWLEDGMENTS
We thank Jorien Favere, Jo De Vrieze, and Maarten De Rijcke for their valuable
feedback on previous versions of the manuscript. We thank Tim Lacoere for his marvelous
graphical designs.

REFERENCES
1. Paau AS, Cowles JR, Oro J. 1977. Flow-microfluorometric analysis of Esch- 5. Chodkowski JL, Shade A. 2017. A synthetic community system for probing
erichia coli, Rhizobium meliloti, and Rhizobium japonicum at different microbial interactions driven by exometabolites. mSystems 2:e00129-17.
stages of the growth cycle. Can J Microbiol 23:1165–1169. https://doi https://doi.org/10.1128/mSystems.00129-17.
.org/10.1139/m77-175. 6. Van Nevel S, Koetzsch S, Proctor CR, Besmer MD, Prest EI, Vrouwenvelder
2. Hutter KJ, Eipel HE. 1979. Microbial determinations by flow cytometry. J JS, Knezev A, Boon N, Hammes F. 2017. Flow cytometric bacterial cell
Gen Microbiol 113:369–375. https://doi.org/10.1099/00221287-113-2-369. counts challenge conventional heterotrophic plate counts for routine
3. Wang Y, Hammes F, De Roy K, Verstraete W, Boon N. 2010. Past, present microbiological drinking water monitoring. Water Res 113:191–206.
and future applications of flow cytometry in aquatic microbiology. Trends https://doi.org/10.1016/j.watres.2017.01.065.
Biotechnol 28:416–424. https://doi.org/10.1016/j.tibtech.2010.04.006. 7. Vandeputte D, Kathagen G, D'hoe K, Vieira-Silva S, Valles-Colomer M, Sabino
4. Frossard A, Hammes F, Gessner MO. 2016. Flow cytometric assessment J, Wang J, Tito RY, De Commer L, Darzi Y, Vermeire S, Falony G, Raes J. 2017.
of bacterial abundance in soils, sediments and sludge. Front Microbiol Quantitative microbiome profiling links gut community variation to micro-
7:903. https://doi.org/10.3389/fmicb.2016.00903. bial load. Nature 551:507–511. https://doi.org/10.1038/nature24460.

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 8

Minireview

8. Gryp T, De Paepe K, Vanholder R, Kerckhof FM, Van Biesen W, Van de Wiele 28. Duquenoy A, Bellais S, Gasc C, Schwintner C, Dore J, Thomas V. 2020.
T, Verbeke F, Speeckaert M, Joossens M, Couttenye MM, Vaneechoutte M, Assessment of Gram- and viability-staining methods for quantifying bacte-
Glorieux G. 2020. Gut microbiota generation of protein-bound uremic tox- rial community dynamics using flow cytometry. Front Microbiol 11:1469.
ins and related metabolites is not altered at different stages of chronic kid- https://doi.org/10.3389/fmicb.2020.01469.
ney disease. Kidney Int 97:1230–1242. https://doi.org/10.1016/j.kint.2020.01 29. Spidlen J, Moore W, Parks D, Goldberg M, Bray C, Bierre P, Gorombey P,
.028. Hyun B, Hubbard M, Lange S, Lefebvre R, Leif R, Novo D, Ostruszka L,
9. Waite AJ, Shou W. 2012. Adaptation to a new environment allows coop- Treister A, Wood J, Murphy RF, Roederer M, Sudar D, Zigon R, Brinkman
erators to purge cheaters stochastically. Proc Natl Acad Sci U S A RR. 2010. Data file standard for flow cytometry, version FCS 3.1. Cytome-
109:19079–19086. https://doi.org/10.1073/pnas.1210190109. try A 77A:97–100. https://doi.org/10.1002/cyto.a.20825.
10. Bleichrodt RJ, Read ND. 2019. Flow cytometry and FACS applied to fila- 30. Hahne F, LeMeur N, Brinkman RR, Ellis B, Haaland P, Sarkar D, Spidlen J,
mentous fungi. Fungal Biol Rev 33:1–15. https://doi.org/10.1016/j.fbr Strain E, Gentleman R. 2009. flowCore: a Bioconductor package for high
.2018.06.001. throughput flow cytometry. BMC Bioinformatics 10:106. https://doi.org/
11. Brussaard CP. 2004. Optimization of procedures for counting viruses by 10.1186/1471-2105-10-106.
flow cytometry. Appl Environ Microbiol 70:1506–1513. https://doi.org/10 31. Arioli S, Guglielmetti S, Amalfitano S, Viti C, Marchi E, Decorosi F,
.1128/AEM.70.3.1506-1513.2004. Giovannetti L, Mora D. 2014. Characterization of tetA-like gene encoding
12. Props R, Monsieurs P, Mysara M, Clement L, Boon N. 2016. Measuring the for a major facilitator superfamily efflux pump in Streptococcus thermo-
biodiversity of microbial communities by flow cytometry. Methods Ecol philus. FEMS Microbiol Lett 355:61–70. https://doi.org/10.1111/1574
Evol 7:1376–1385. https://doi.org/10.1111/2041-210X.12607. -6968.12449.
13. Hammes F, Broger T, Weilenmann HU, Vital M, Helbing J, Bosshart U, 32. Anvarian AHP, Cao Y, Srikumar S, Fanning S, Jordan K. 2016. Flow cyto-
Huber P, Peter Odermatt R, Sonnleitner B. 2012. Development and labo- metric and 16S sequencing methodologies for monitoring the physio-
ratory-scale testing of a fully automated online flow cytometer for drink- logical status of the microbiome in powdered infant formula production.
ing water analysis. Cytometry A 81A:508–516. https://doi.org/10.1002/ Front Microbiol 7:968. https://doi.org/10.3389/fmicb.2016.00968.
cyto.a.22048. 33. Conacher CG, Naidoo-Blassoples RK, Rossouw D, Bauer FF. 2020. Real-
14. Besmer MD, Hammes F. 2016. Short-term microbial dynamics in a drinking time monitoring of population dynamics and physical interactions in a
water plant treating groundwater with occasional high microbial loads. synthetic yeast ecosystem by use of multicolour flow cytometry. Appl
Water Res 107:11–18. https://doi.org/10.1016/j.watres.2016.10.041. Microbiol Biotechnol 104:5547–5562. https://doi.org/10.1007/s00253
15. Props R, Rubbens P, Besmer M, Buysschaert B, Sigrist J, Weilenmann H, -020-10607-x.
Waegeman W, Boon N, Hammes F. 2018. Detection of microbial distur- 34. Vergine P, Amalfitano S, Salerno C, Berardi G, Pollice A. 2020. Reuse of
bances in a drinking water microbial community through continuous ac- ultrafiltered effluents for crop irrigation: on-site flow cytometry unveiled
quisition and advanced analysis of flow cytometry data. Water Res microbial removal patterns across a full-scale tertiary treatment. Sci Total
145:73–82. https://doi.org/10.1016/j.watres.2018.08.013. Environ 718:137298. https://doi.org/10.1016/j.scitotenv.2020.137298.
16. Müller S, Nebe-von Caron G. 2010. Functional single-cell analyses: flow 35. Parks DR, Roederer M, Moore W. 2006. A new “Logicle” display method
cytometry and cell sorting of microbial populations and communities. avoids deceptive effects of logarithmic scaling for low signals and com-
FEMS Microbiol Rev 34:554–587. https://doi.org/10.1111/j.1574-6976.2010 pensated data. Cytometry A 69A:541–551. https://doi.org/10.1002/cyto.a
.00214.x. .20258.
17. Hatzenpichler R, Krukenberg V, Spietz RL, Jay ZJ. 2020. Next-generation 36. Herzenberg L, Tung J, Moore W, Herzenberg L, Parks DR. 2006. Interpret-
physiology approaches to study microbiome function at single cell level. ing flow cytometry data: a guide for the perplexed. Nat Immunol
Nat Rev Microbiol 18:241–256. https://doi.org/10.1038/s41579-020-0323-1. 7:681–685. https://doi.org/10.1038/ni0706-681.
18. Beller A, Kruglov A, Durek P, Goetze V, Werner K, Heinz GA, Ninnemann J, 37. Finak G, Perez JM, Weng A, Gottardo R. 2010. Optimizing transformations
Lehmann K, Maier R, Hoffmann U, Riedel R, Heiking K, Zimmermann J, for automated, high throughput analysis of flow cytometry data. BMC
Siegmund B, Mashreghi M, Radbruch A, Chang H. 2020. Specific micro- Bioinformatics 11:546. https://doi.org/10.1186/1471-2105-11-546.
biota enhances intestinal IgA levels by inducing TGF- b in T follicular 38. Sarkar D, Le Meur N, Gentleman R. 2008. Using flowViz to visualize flow

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

helper cells of Peyer’s patches in mice. Eur J Immunol 50:783–794. cytometry data. Bioinformatics 24:878–879. https://doi.org/10.1093/
https://doi.org/10.1002/eji.201948474. bioinformatics/btn021.
19. O'Neill K, Aghaeepour N, Spidlen J, Brinkman R. 2013. Flow cytometry 39. Van P, Jiang W, Gottardo R, Finak G. 2018. ggCyto: next generation open-
bioinformatics. PLoS Comput Biol 9:e1003365. https://doi.org/10.1371/ source visualization software for cytometry. Bioinformatics 34:3951–3953.
journal.pcbi.1003365. https://doi.org/10.1093/bioinformatics/bty441.
20. Saeys Y, Van Gassen S, Lambrecht BN. 2016. Computational flow cytome- 40. Brown MR, Hands CL, Coello-Garcia T, Sani BS, Ott AIG, Smith SJ,
try: helping to make sense of high-dimensional immunology data. Nat Davenport RJ. 2019. A flow cytometry method for bacterial quantifica-
Rev Immunol 16:449–462. https://doi.org/10.1038/nri.2016.56. tion and biomass estimates in activated sludge. J Microbiol Methods
21. Quixabeira VBL, Nabout JC, Rodrigues FM. 2010. Trends in genetic litera- 160:73–83. https://doi.org/10.1016/j.mimet.2019.03.022.
ture with the use of flow cytometry. Cytometry A 77A:207–210. https:// 41. Fletez-Brant K, Spidlen J, Brinkman RR, Roederer M, Chattopadhyay PK.
doi.org/10.1002/cyto.a.20837. 2016. flowClean: automated identification and removal of fluorescence
22. Mair F, Prlic M. 2018. OMIP 044: 28 color immunophenotyping of the anomalies in flow cytometry data. Cytometry A 89A:461–471. https://doi
human dendritic cell compartment. Cytometry A 93A:402–405. https:// .org/10.1002/cyto.a.22837.
doi.org/10.1002/cyto.a.23331. 42. Monaco G, Chen H, Poidinger M, Chen J, de Magalhães JP, Larbi A. 2016.
23. Buysschaert B, Byloos B, Leys N, Van Houdt R, Boon N. 2016. Reevaluating flowAI: automatic and interactive anomaly discerning tools for flow
multicolor flow cytometry to assess microbial viability. Appl Microbiol cytometry data. Bioinformatics 32:2473–2480. https://doi.org/10.1093/
Biotechnol 100:9037–9051. https://doi.org/10.1007/s00253-016-7837-5. bioinformatics/btw191.
24. Léonard L, Chibane LB, Bouhedda BO, Degraeve P, Oulahal N. 2016. 43. Meskas J, Wang S, Brinkman RR. 2020. flowCut—an R package for precise
Recent advances on multi-parameter flow cytometry to characterize anti- and accurate automated removal of outlier events and flagging of files
microbial treatments. Front Microbiol 7:1225. https://doi.org/10.3389/ based on time versus fluorescence analysis. bioRxiv https://doi.org/10
fmicb.2016.01225. .1101/2020.04.23.058545.
25. Koch C, Müller S. 2018. Personalized microbiome dynamics—cytometric 44. Hammes F, Egli T. 2010. Cytometric methods for measuring bacteria in water:
fingerprints for routine diagnostics. Mol Aspects Med 59:123–134. advantages, pitfalls and applications. Anal Bioanal Chem 397:1083–1095.
https://doi.org/10.1016/j.mam.2017.06.005. https://doi.org/10.1007/s00216-010-3646-3.
26. Hatzenpichler R, Connon SA, Goudeau D, Malmstrom RR, Woyke T, 45. Fragoso GM, Poulton AJ, Pratt NJ, Johnsen G, Purdie DA. 2019. Trait-
Orphan VJ. 2016. Visualizing in situ translational activity for identifying based analysis of subpolar North Atlantic phytoplankton and plastidic
and sorting slow-growing archaeal-bacterial consortia. Proc Natl Acad ciliate communities using automated flow cytometer. Limnol Oceanogr
Sci U S A 113:E4069–E4078. https://doi.org/10.1073/pnas.1603757113. 64:1763–1778. https://doi.org/10.1002/lno.11189.
27. Barbesti S, Citterio S, Labra M, Baroni MD, Neri MG, Sgorbati S. 2000. Two 46. Proctor CR, Besmer MD, Langenegger T, Beck K, Walser JC, Ackermann
and three-color fluorescence flow cytometric analysis of immunoidenti- M, Bürgmann H, Hammes F. 2018. Phylogenetic clustering of small low
fied viable bacteria. Cytometry 40:214–218. https://doi.org/10.1002/1097 nucleic acid-content bacteria across diverse freshwater ecosystems.
-0320(20000701)40:3,214::AID-CYTO6.3.0.CO;2-M. ISME J 12:1344–1359. https://doi.org/10.1038/s41396-018-0070-8.

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 9

Minireview

47. van der Maaten L, Hinton G. 2008. Visualizing data using t-SNE. J Mach 68. Aghaeepour N, Chattopadhyay P, Chikina M, Dhaene T, Van Gassen S,
Learn Res 9:2579–2605. Kursa M, Lambrecht BN, Malek M, Mclachlan GJ, Qian Y, Qiu P, Saeys Y,
48. McInnes L, Healy J, Saul N, Großberger L. 2018. UMAP: uniform manifold Stanton R, Tong D, Vens C, Walkowiak S, Wang K, Finak G, Gottardo R,
approximation and projection. J Open Source Softw 3:861. https://doi Mosmann T, Nolan GP, Scheuermann RH, Brinkman RR. 2016. A bench-
.org/10.21105/joss.00861. mark for evaluation of algorithms for identification of cellular correlates
49. Aghaeepour N, Finak G, Hoos H, Mosmann TR, Brinkman R, Gottardo R, of clinical outcomes. Cytometry A 89A:16–21. https://doi.org/10.1002/
Scheuermann RH, DREAM Consortium. 2013. Critical assessment of auto- cyto.a.22732.
mated flow cytometry data analysis techniques. Nat Methods 10:228–238. 69. Kotecha N, Krutzik PO, Irish JM. 2010. Web-based analysis and publication
https://doi.org/10.1038/nmeth.2365. of flow cytometry experiments. Curr Protoc Cytom 53:10.17.1–10.17.24.
50. Weber LM, Robinson MD. 2016. Comparison of clustering methods for https://doi.org/10.1002/0471142956.cy1017s53.
high-dimensional single-cell flow and mass cytometry data. Cytometry A 70. Spidlen J, Breuer K, Rosenberg C, Kotecha N, Brinkman RR. 2012. Flow-
89A:1084–1096. https://doi.org/10.1002/cyto.a.23030. Repository: a resource of annotated flow cytometry datasets associated
51. Liu X, Song W, Wong BY, Zhang T, Yu S, Lin GN, Ding X. 2019. A compari- with peer-reviewed publications. Cytometry A 81A:727–731. https://
son framework and guideline of clustering methods for mass cytometry doi.org/10.1002/cyto.a.22106.
data. Genome Biol 20:297. https://doi.org/10.1186/s13059-019-1917-7. 71. Lee JA, Spidlen J, Boyce K, Cai J, Crosbie N, Dalphin M, Furlong J, Gasparetto
52. Van Gassen S, Callebaut B, Van Helden MJ, Lambrecht BN, Demeester P,
M, Goldberg M, Goralczyk EM, Hyun B, Jansen K, Kollmann T, Kong M, Leif R,
Dhaene T, Saeys Y. 2015. FlowSOM: using self-organizing maps for visu-
McWeeney S, Moloshok TD, Moore W, Nolan G, Nolan J, Nikolich-Zugich J,
alization and interpretation of cytometry data. Cytometry A 87A:636–645.
Parrish D, Purcell B, Qian Y, Selvaraj B, Smith C, Tchuvatkina O, Wertheimer
https://doi.org/10.1002/cyto.a.22625.
A, Wilkinson P, Wilson C, Wood J, Zigon R, Scheuermann RH, Brinkman RR,
53. Levine JH, Simonds EF, Bendall SC, Davis KL, Amir E-a. D, Tadmor MD,
Litvin O, Fienberg HG, Jager A, Zunder ER, Finck R, Gedman AL, Radtke I, International Society for Advancement of Cytometry Data Standards Task
Downing JR, Pe'er D, Nolan GP. 2015. Data-driven phenotypic dissection Force. 2008. MIFlowCyt: the minimum information about a flow cytometry
of AML reveals progenitor-like cells that correlate with prognosis. Cell experiment. Cytometry A 73A:926–930. https://doi.org/10.1002/cyto.a.20623.
162:184–197. https://doi.org/10.1016/j.cell.2015.05.047. 72. Chan S, Pullerits K, Riechelmann J, Persson KM, Rådström P, Paul CJ.
54. Ludwig J, zu Siederdissen CH, Liu Z, Stadler PF, Müller S. 2019. flowEMMi: 2018. Monitoring biofilm function in new and matured full-scale slow
an automated model-based clustering tool for microbial cytometric data. sand filters using flow cytometric histogram image comparison (CHIC).
BMC Bioinformatics 20:643. https://doi.org/10.1186/s12859-019-3152-3. Water Res 138:27–36. https://doi.org/10.1016/j.watres.2018.03.032.
55. Rajwa B, Venkatapathi M, Ragheb K, Banada PP, Hirleman ED, Lary T, 73. Duveau F, Hodgins-Davis A, Metzger BPH, Yang B, Tryban S, Walker EA,
Robinson JP. 2008. Automated classification of bacterial particles in flow Lybrook T, Wittkopp PJ. 2018. Fitness effects of altering gene expression
by multiangle scatter measurement and support vector machine classi- noise in Saccharomyces cerevisiae. Elife 7:e37272. https://doi.org/10
fier. Cytometry A 73A:369–379. https://doi.org/10.1002/cyto.a.20515. .7554/eLife.37272.
56. Pereira GC, Ebecken NFF. 2011. Combining in situ flow cytometry and ar- 74. Karava M, Bracharz F, Kabisch J. 2019. Quantification and isolation of Ba-
tificial neural networks for aquatic systems monitoring. Expert Syst Appl cillus subtilis spores using cell sorting and automated gating. PLoS One
38:9626–9632. https://doi.org/10.1016/j.eswa.2011.01.140. 14:e0219892-15. https://doi.org/10.1371/journal.pone.0219892.
57. Rubbens P, Props R, Boon N, Waegeman W. 2017. Flow cytometric sin- 75. Rubbens P, Props R, Garcia-Timermans C, Boon N, Waegeman W. 2017.
gle-cell identification of populations in synthetic bacterial communities. Stripping flow cytometry: How many detectors do we need for bacterial
PLoS One 12:e0169754. https://doi.org/10.1371/journal.pone.0169754. identification? Cytometry A 91A:1184–1191. https://doi.org/10.1002/cyto
58. Duygan BDÖ, Hadadi N, Babu AF, Seyfried M, van der Meer JR. 2020. .a.23284.
Rapid detection of microbiota cell type diversity using machine-learned 76. Buysschaert B, Kerckhof FM, Vandamme P, De Baets B, Boon N. 2018.
classification of flow cytometry data. Commun Biol 3:379. https://doi Flow cytometric fingerprinting for microbial strain discrimination and
.org/10.1038/s42003-020-1106-y. physiological characterization. Cytometry A 93A:201–212. https://doi.org/
59. García-Timermans C, Rubbens P, Heyse J, Kerckhof FM, Props R, Skirtach 10.1002/cyto.a.23302.

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

AG, Waegeman W, Boon N. 2020. Discriminating bacterial phenotypes at 77. Hahne F, Khodabakhshi AH, Bashashati A, Wong CJ, Gascoyne RD, Weng
the population and single-cell level: a comparison of flow cytometry and AP, Seyfert-Margolis V, Bourcier K, Asare A, Lumley T, Gentleman R,
Raman spectroscopy fingerprinting. Cytometry A 97A:713–714. https:// Brinkman RR. 2010. Per-channel basis normalization methods for flow
doi.org/10.1002/cyto.a.23952. cytometry data. Cytometry A 77A:121–131. https://doi.org/10.1002/cyto
60. Koch C, Fetzer I, Schmidt T, Harms H, Müller S. 2013. Monitoring func- .a.20823.
tions in managed microbial systems by cytometric bar coding. Environ 78. Jahn M, Seifert J, Hübschmann T, Von Bergen M, Harms H, Müller S. 2013.
Sci Technol 47:1753–1760. https://doi.org/10.1021/es3041048. Comparison of preservation methods for bacterial cells in cytomics and
61. Koch C, Fetzer I, Harms H, Müller S. 2013. CHIC—an automated approach proteomics. J Integr OMICS 3:25–33. https://doi.org/10.5584/jiomics.v3i1
for the detection of dynamic variations in complex microbial commun-
.115.
ities. Cytometry A 83A:561–567. https://doi.org/10.1002/cyto.a.22286.
79. Vazquez-Albacete D, Cavaleiro AM, Christensen U, Seppälä S, Møller
62. Wanderley BMS, A Araújo DS, Quiroga MV, Amado AM, Neto ADD,
BL, Nørholm MHH. 2017. An expression tag toolbox for microbial pro-
Sarmento H, Metz SD, Unrein F. 2019. flowDiv: a new pipeline for analyz-
duction of membrane bound plant cytochromes P450. Biotechnol
ing flow cytometric diversity. BMC Bioinformatics 20:274. https://doi
Bioeng 114:751–760. https://doi.org/10.1002/bit.26203.
.org/10.1186/s12859-019-2787-4.
80. Reiss J, Perkins DM, Fussmann KE, Krause S, Canhoto C, Romeijn P,
63. Roederer M, Moore W, Treister A, Hardy RR, Herzenberg LA. 2001. Proba-
Robertson AL. 2019. Groundwater flooding: ecosystem structure follow-
bility binning comparison: a metric for quantitating multivariate distri-
bution differences. Cytometry 45:47–55. https://doi.org/10.1002/1097 ing an extreme recharge event. Sci Total Environ 652:1252–1260. https://
-0320(20010901)45:1,47::AID-CYTO1143.3.0.CO;2-A. doi.org/10.1016/j.scitotenv.2018.10.216.
64. Rogers WT, Holyst HA. 2009. FlowFP: a Bioconductor package for finger- 81. Amir E-A. D, Davis KL, Tadmor MD, Simonds EF, Levine JH, Bendall SC,
printing flow cytometric data. Adv Bioinform 2009:193947. https://doi Shenfeld DK, Krishnaswamy S, Nolan GP, Pe'er D. 2013. viSNE enables vis-
.org/10.1155/2009/193947. ualization of high dimensional single-cell data and reveals phenotypic
65. Huang TH, Ning X, Wang X, Murthy N, Tzeng YL, Dickson RM. 2015. heterogeneity of leukemia. Nat Biotechnol 31:545–552. https://doi.org/
Rapid cytometric antibiotic susceptibility testing utilizing adaptive 10.1038/nbt.2594.
multidimensional statistical metrics. Anal Chem 87:1941–1949. 82. Sgier L, Freimann R, Zupanic A, Kroll A. 2016. Flow cytometry combined
https://doi.org/10.1021/ac504241x. with viSNE for the analysis of microbial biofilms and detection of micro-
66. Amalfitano S, Fazi S, Ejarque E, Freixa A, Romaní AM, Butturini A. 2018. plastics. Nat Commun 7:11587. https://doi.org/10.1038/ncomms11587.
Deconvolution model to resolve cytometric microbial community pat- 83. Becht E, McInnes L, Healy J, Dutertre CA, Kwok IWH, Ng LG, Ginhoux F,
terns in flowing waters. Cytometry A 93A:194–200. https://doi.org/10 Newell EW. 2019. Dimensionality reduction for visualizing single-cell data
.1002/cyto.a.23304. using UMAP. Nat Biotechnol 37:38–44. https://doi.org/10.1038/nbt.4314.
67. Rubbens P, Props R, Kerckhof FM, Boon N, Waegeman W. PhenoGMM: 84. Lo K, Hahne F, Brinkman RR, Gottardo R. 2009. flowClust: a Bioconductor
Gaussian mixture modelling of cytometry data quantifies changes in mi- package for automated gating of flow cytometry data. BMC Bioinfor-
crobial community structure. mSphere, in press. matics 10:145. https://doi.org/10.1186/1471-2105-10-145.

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 10

Minireview

85. García FC, López-Urrutia Á, Morán XAG. 2014. Automated clustering of 102. Ramette A. 2007. Multivariate analyses in microbial ecology. FEMS Micro-
heterotrophic bacterioplankton in flow cytometry data. Aquat Microb biol Ecol 62:142–160. https://doi.org/10.1111/j.1574-6941.2007.00375.x.
Ecol 72:175–185. https://doi.org/10.3354/ame01691. 103. Paliy O, Shankar V. 2016. Application of multivariate statistical techni-
86. Ge Y, Sealfon SC. 2012. flowPeaks: a fast unsupervised clustering for flow ques in microbial ecology. Mol Ecol 25:1032–1057. https://doi.org/10
cytometry data via K-means and density peak finding. Bioinformatics .1111/mec.13536.
28:2052–2058. https://doi.org/10.1093/bioinformatics/bts300. 104. Fontana S, Petchey OL, Pomati F. 2016. Individual level trait diversity
87. Bervoets I, Van Brempt M, Van Nerom K, Van Hove B, Maertens J, De Mey concepts and indices to comprehensively describe community change
M, Charlier D. 2018. A sigma factor toolbox for orthogonal gene expres- in multidimensional trait space. Funct Ecol 30:808–818. https://doi.org/
sion in Escherichia coli. Nucleic Acids Res 46:2133–2144. https://doi.org/ 10.1111/1365-2435.12551.
10.1093/nar/gky010. 105. Liu Z, Cichocki N, Bonk F, Günther S, Schattenberg F, Harms H, Centler F,
88. Thomas MK, Fontana S, Reyes M, Pomati F. 2018. Quantifying cell den- Müller S. 2018. Ecological stability properties of microbial communities
sities and biovolumes of phytoplankton communities and functional assessed by flow cytometry. mSphere 3:e00564-17. https://doi.org/10
groups using scanning flow cytometry, machine learning and unsuper- .1128/mSphere.00564-17.
vised clustering. PLoS One 13:e0196225. https://doi.org/10.1371/journal 106. Christensen H, Bakken LR, Olsen RA. 1993. Soil bacterial DNA and bio-
.pone.0196225.
volume profiles measured by flow-cytometry. FEMS Microbiol Lett
89. Malek M, Taghiyar MJ, Chong L, Finak G, Gottardo R, Brinkman RR.
102:129–140. https://doi.org/10.1111/j.1574-6968.1993.tb05804.x.
2015. flowDensity: reproducing manual gating of flow cytometry data
107. Bouvier T, Troussellier M, Anzil A, Courties C, Servais P. 2001. Using light
by automated density-based cell population identification. Bioinfor-
scatter signal to estimate bacterial biovolume by flow cytometry. Cytome-
matics 31:606–607. https://doi.org/10.1093/bioinformatics/btu677.
try 44:188–194. https://doi.org/10.1002/1097-0320(20010701)44:3,188::
90. Ribalet F, Berthiaume C, Hynes A, Swalwell J, Carlson M, Clayton S,
Hennon G, Poirier C, Shimabukuro E, White A, Armbrust EV. 2019. Sea- AID-CYTO1111.3.0.CO;2-C.
Flow data v1, high-resolution abundance, size and biomass of small phy- 108. Felip M, Andreatta S, Sommaruga R, Straskrábová V, Catalan J, 2007.
toplankton in the North Pacific. Sci Data 6:277. https://doi.org/10.1038/ Suitability of flow cytometry for estimating bacterial biovolume in natu-
s41597-019-0292-2. ral plankton samples: comparison with microscopy data. Appl Environ
91. Koch C, Harnisch F, Schröder U, Müller S. 2014. Cytometric fingerprints: Microbiol 73:4508–4514. https://doi.org/10.1128/AEM.00733-07.
evaluation of new tools for analyzing microbial community dynamics. 109. Jang J, Hendriksen NB, Jakobsen HH, Gosewinkel U. 2018. Application of
Front Microbiol 5:273. https://doi.org/10.3389/fmicb.2014.00273. Cytosense flow cytometer for the analysis of airborne bacteria collected
92. Menyhárt L, Nagy S, Lepossa A. 2018. Rapid analysis of photoautotroph mi- by a high volume impingement sampler. J Microbiol Methods 154:63–72.
crobial communities in soils by flow cytometric barcoding and fingerprinting. https://doi.org/10.1016/j.mimet.2018.10.012.
Appl Soil Ecol 130:237–240. https://doi.org/10.1016/j.apsoil.2018.06.013. 110. Singh A, Yelverton CJ, Barnard TG. 2019. Rapid quantification of the total
93. van Gelder S, Röhrig N, Schattenberg F, Cichocki N, Schumann J, Schmalz viable bacterial population on human hands using flow cytometry with
G, Haak R, Ziebolz D, Müller S. 2018. A cytometric approach to follow varia- SYBR Green I. Cytometry B Clin Cytom 96:397–403. https://doi.org/10
tion and dynamics of the salivary microbiota. Methods 134-135:67–79. .1002/cyto.b.21776.
https://doi.org/10.1016/j.ymeth.2017.08.009. 111. Bowman JS, Amaral-Zettler LA, J Rich J, M Luria C, Ducklow HW. 2017.
94. De Roy K, Clement L, Thas O, Wang Y, Boon N. 2012. Flow cytometry for Bacterial community segmentation facilitates the prediction of ecosys-
fast microbial community fingerprinting. Water Res 46:907–919. https:// tem function along the coast of the western Antarctic Peninsula. ISME J
doi.org/10.1016/j.watres.2011.11.076. 11:1460–1471. https://doi.org/10.1038/ismej.2016.204.
95. Kinet R, Dzaomuho P, Baert J, Taminiau B, Daube G, Nezer C, Brostaux Y, 112. Schleich C, Chan S, Pullerits K, Besmer MD, Paul CJ, Rådström P, Keucken
Nguyen F, Dumont G, Thonart P, Delvigne F. 2016. Flow cytometry commu- A. 2019. Mapping dynamics of bacterial communities in a full-scale
nity fingerprinting and amplicon sequencing for the assessment of landfill drinking water distribution system using flow cytometry. Water 11:2137.
leachate cellulolytic bioaugmentation. Bioresour Technol 214:450–459. https://doi.org/10.3390/w11102137.
https://doi.org/10.1016/j.biortech.2016.04.131. 113. Huang TH, Tzeng YL, Dickson RM. 2018. FAST: rapid determinations of
96. Favere J, Buysschaert B, Boon N, De Gusseme B. 2020. Online microbial

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

antibiotic susceptibility phenotypes using label-free cytometry. Cyto-
fingerprinting for quality management of drinking water: full-scale event metry A 93A:639–648. https://doi.org/10.1002/cyto.a.23370.
detection. Water Res 170:115353. https://doi.org/10.1016/j.watres.2019 114. Props R, Kerckhof FM, Rubbens P, De Vrieze J, Hernandez Sanabria E,
.115353. Waegeman W, Monsieurs P, Hammes F, Boon N. 2017. Absolute quantifi-
97. Coggins LX, Larma I, Hinchliffe A, Props R, Ghadouani A. 2020. Flow cation of microbial taxon abundances. ISME J 11:584–587. https://doi
cytometry for rapid characterisation of microbial community dynamics
.org/10.1038/ismej.2016.117.
in waste stabilisation ponds. Water Res 169:115243. https://doi.org/10
115. Props R, Schmidt ML, Heyse J, Vanderploeg HA, Boon N, Denef VJ. 2018.
.1016/j.watres.2019.115243.
Flow cytometric monitoring of bacterioplankton phenotypic diversity pre-
98. De Rudder C, Calatayud Arroyo M, Lebeer S, Van de Wiele T. 2020. Dual and
dicts high population-specific feeding rates by invasive dreissenid mussels.
triple epithelial coculture model systems with donor-derived microbiota
Environ Microbiol 20:521–534. https://doi.org/10.1111/1462-2920.13953.
and THP-1 macrophages to mimic host-microbe interactions in the human
116. García FC, Alonso-Sáez L, Morán XAG, López-Urrutia Á. 2015. Seasonality
sinonasal cavities. mSphere 5:e00916-19. https://doi.org/10.1128/mSphere
.00916-19. in molecular and cytometric diversity of marine bacterioplankton: the
99. Rubbens P, Props R, Kerckhof FM, Boon N, Waegeman W. 2021. Cytomet- re-shuffling of bacterial taxa by vertical mixing. Environ Microbiol
ric fingerprints of gut microbiota predict Crohn's disease state. ISME J 17:4133–4142. https://doi.org/10.1111/1462-2920.12984.
15:354–358. https://doi.org/10.1038/s41396-020-00762-4. 117. Jahn M, Seifert J, von Bergen M, Schmid A, Bühler B, Müller S. 2013. Sub-
100. Fodelianakis S, Lorz A, Valenzuela-Cuevas A, Barozzi A, Booth JM, population-proteomics in prokaryotic populations. Curr Opin Biotechnol
Daffonchio D. 2019. Dispersal homogenizes communities via immigration 24:79–87. https://doi.org/10.1016/j.copbio.2012.10.017.
even at low rates in a simplified synthetic bacterial metacommunity. Nat 118. Stepanauskas R, Fergusson EA, Brown J, Poulton NJ, Tupper B, Labonté
Commun 10:1314. https://doi.org/10.1038/s41467-019-09306-7. JM, Becraft ED, Brown JM, Pachiadaki MG, Povilaitis T, Thompson BP,
101. Koch C, Günther S, Desta AF, Hübschmann T, Müller S. 2013. Cytometric Mascena CJ, Bellows WK, Lubys A. 2017. Improved genome recovery
fingerprinting for analyzing microbial intracommunity structure varia- and integrated cell-size analyses of individual uncultured microbial cells
tion and identifying subcommunity function. Nat Protoc 8:190–202. and viral particles. Nat Commun 8:84. https://doi.org/10.1038/s41467
https://doi.org/10.1038/nprot.2012.149. -017-00128-z.

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 11

Minireview

Peter Rubbens is a postdoctoral researcher

at the Flanders Marine Institute (VLIZ) in
Ostend, Belgium. He obtained his Ph.D. at
Ghent University in 2019. During his doc-
toral study, he investigated and specialized
in machine learning approaches for micro-
bial flow cytometry data, in addition to
other single-cell technologies. Currently, his
research has shifted to a quantitative study
of marine microbial communities and their
dynamics in response to a changing envi-
ronment. In addition, he is interested in applications of machine learning
for microbial ecology.

Ruben Props is a postdoctoral researcher

at the Center for Microbial Ecology and
Technology (CMET; Ghent University). He
obtained his Ph.D. at Ghent University in
2018. During his doctoral studies, he stud-
ied the microbial ecology of both engi-
neered and freshwater ecosystems. He car-
ried out part of his Ph.D. research at the
University of Michigan under the supervi-
sion of Vincent Denef, during which he
studied strain-level variations in abundant
bacterioplankton and the effects of invasive mussel species on bacterio-
plankton composition and function. Currently, he is investigating the
phylogenomic basis of phenotypic properties of microbes using single-
cell analysis techniques.

Downloaded from https://journals.asm.org/journal/msystems on 25 May 2023 by 217.69.247.113.

January/February 2021 Volume 6 Issue 1 e00895-20 msystems.asm.org 12