LAUSR

WHEN late at night (as I imagine it), Howard Shapiro finally came out of his laboratory, where he spent the whole day, he was carrying a massive ring binder in his hands to present the Laws of Flow Cytometry (https://bitesizebio.com/22533/shapiros-laws-offlow-cytometry/) to the expectant community. Today, I want to remind you of his seventh law which says: “No Data Analysis Technique Can Make Good Data out of Bad Data.” As Rachel Walker excellently paraphrased (https:// bitesizebio.com/22533/shapiros-laws-of-flow-cytometry/): “This is arguably Shapiro’s most important law. If we refer back to the article ‘Crap in Crap out: Flushing Out the Out The Problems in Your Flow Cytometry Data’, if you have a bad sample, bad staining, etc., running those results through an analysis package is not suddenly going to give you a fabulous Nobel Prize winning result. If your cells don’t look right when running them on the cytometer, they are not going to suddenly look good in post-acquisition data analysis. This links to the Zero law – there is no magic.” There is a frequent view that the more different markers you use to define cell types by high-dimensional fluorescent and mass cytometry techniques, the better cell populations can be identified. High-dimensional multicolor cell analysis is now common sense. It is done either by combining several medium polychromatic panels like in the work of Pitoiset et al. (this issue, p. 793) who developed a series of 10-color panels for comprehensive immunophenotyping, or by combining as many colors as possible or limited by the number of detectors in an instrument. The race is on and last year’s top ranking OMIP with 16 colors (1) was already topped by 28-colors this April (2) catching up with mass cytometry’s dimensionality. Cell clustering methods help to (automatically) identify unequivocally each cell subset and support discovery of new ones, hitherto undetected (3). Although this perception is basically not incorrect, Shapiro’s seventh law still applies: nonoptimized cell preparation and labeling lead to mediocre quality results. That is why this month’s Editor’s Choice selection is on the aspect of clustering analysis of highly complex flow cytometry data and its biases. Mazza et al. (this issue, p. 785) focused on 27-parameter flow cytometry data analysis by tSNE. They could identify fluorescence spillover and background as sources of variance of their clustering results leading accidentally to erroneous “new” cell populations. The authors conclude that optimized panels with low background and spill-over and further precautions in experimental design help to improve the reliability of cluster analysis. Importantly, critical evaluation of all aspects of cell treatment is pivotal for single cell proteomics (4) or in future maybe also genomics by flow cytometry. Donnenberg et al. (this issue, p. 803) evaluated the impact of trypsin treatment on adherent or tissue cells for flow cytometric surface proteome determination. The authors conclude that carefully controlled trypsinization has relatively minute influence on the antigens tested. In the same line, Formamide, known for being used in in situ hybridization, is since recently also an interesting reagent for quantitative DNA content determination (5). Here, Radicchio et al. (this issue, p. 829) show that Formamide can be applied for combining surface antigen determination with high-quality DNA flow cytometry. For further reading on Nuclear Cytometry, I also recommend the review by Smith et al. (this issue, p. 771). Finally, I want to highlight three studies involving microfluidic cytometry presumably the next generation in flow cytometry en large. Mechanical properties of cells have the