cyCombine allows for robust integration of single-cell cytometry datasets within and across technologies

Abstract

Combining single-cell cytometry datasets increases the analytical flexibility and the statistical power of data analyses. However, in many cases the full potential of co-analyses is not reached due to technical variance between data from different experimental batches. Here, we present cyCombine, a method to robustly integrate cytometry data from different batches, experiments, or even different experimental techniques, such as CITE-seq, flow cytometry, and mass cytometry. We demonstrate that cyCombine maintains the biological variance and the structure of the data, while minimizing the technical variance between datasets. cyCombine does not require technical replicates across datasets, and computation time scales linearly with the number of cells, allowing for integration of massive datasets. Robust, accurate, and scalable integration of cytometry data enables integration of multiple datasets for primary data analyses and the validation of results using public datasets.

Integrated machine learning approaches for flow cytometric quantification of myeloid-derived suppressor cells in acute sepsis

“…Batch effects need to be accounted for when combining results obtained on different experimental days. Our analyses demonstrate an appropriate compensation for batch effects by using the cyCombine tool ( 23 ). CytoNorm is another Flowjo tool that is available for correction of batch effects.…”

“…Given the stepwise collection of blood samples over the study period, batch correction was performed by using cyCombine, a plugin running in the R programming language ( Figure 1C ) ( 23 ). CyCombine integrates cytometry data from different experiments, maintaining the biologic variance and structure of data while minimizing dataset differences due to technical variance ( 23 ).…”

“…Given the stepwise collection of blood samples over the study period, batch correction was performed by using cyCombine, a plugin running in the R programming language ( Figure 1C ) ( 23 ). CyCombine integrates cytometry data from different experiments, maintaining the biologic variance and structure of data while minimizing dataset differences due to technical variance ( 23 ). Supervised clustering was then performed by FlowSOM v3.0.18 (R language plugin), a self-organizing map algorithm that provides visual representation of data by using a minimal spanning tree ( 24 ) ( Figure 1D ).…”

Human immune and gut microbial parameters associated with inter-individual variations in COVID-19 mRNA vaccine-induced immunity

“…Clean-up gates for live single cells and elimination of non-cell signals were manually conducted using the web-plat software, Cytobank (v.9.1). To correct batch effects across CyTOF runs, signal intensities were normalized using cyCombine 46 . Data were analyzed using a previously described R-based pipeline 47 .…”

High-throughput mass cytometry staining for deep phenotyping of human natural killer cells

“…Several algorithms are available to correct the batch effect ( Lo et al., 2022 ; Pedersen et al., 2022 ; Van Gassen et al., 2020 ). To do this, the use of validated and appropriate internal control is required.…”