Abstract
Background
Conventional data analysis of flow cytometry-based basophil activation testing requires repetitive, labor-intensive analysis that hampers efforts to standardize testing for clinical applications. Using an open-source platform, we developed and implemented a programmatic approach to the analysis of the basophil activation test (BAT) by flow cytometry.
Methods
Using the BÜHLMANN FlowCAST® assay, peripheral blood from peanut allergic patients undergoing oral immunotherapy was incubated with peanut allergens (Arah1, Arah2, Arah6, whole peanut extract) and stained with fluorescent antibodies to CCR3 and CD63 for the development of a data-driven programmatic analysis using Bioconductor and R. Basophil identification using clustering and classification was validated using manually gated comparisons in an experimental subset. Reproducibility of CD63 upregulation set on unstimulated or anti-FcERI stimulated basophils was compared.
Results
BAT analysis of 294 experiments was successful in 91.5% using the above approach, with a total of 7,166 individual basophil activation tests from 269 experiments. We estimate this represents a net saving of 1340 minutes of labor by a skilled operator. Medium-based gating correlated to respective manual gating more closely than anti-FcERI based gating (R=0.96 vs. R=0.84, p<0.001). Only 2% of the basophil activation results were significantly different from manual gating. Quality measures of the experiments and other measures of basophil activation were also provided by the analysis.
Conclusions
We present a novel data-driven flow cytometric platform for the analysis of clinical basophil activation testing, providing a high throughput objective approach to basophil activation analysis.
Keywords: computational analysis, basophil granulocytes, allergy
Introduction
Reproducible research is increasingly important in this era of globalized science as a way for scientists to transparently share data analysis. Conventional data analysis of flow cytometry data is repetitive and labor-intensive, lacks transparency and is susceptible to bias. Particularly in the context of large clinical trials, in which data acquisition can occur longitudinally over a long time period, sometimes on different instruments and at multiple centers,(1, 2) the ability to use a programmatic gating strategy not only is more transparent and reproducible but also facilitates better quality control that improve interpretation of the data(3–5).
Measurement of antigen-induced basophil activation ex vivo has been shown to be a promising diagnostic test in many allergic disorders.(6–8) Suppression of basophil activation has been associated with immunotherapy to both environmental(9, 10) and food allergens.(1, 11–14) Historically, identification of rare cells such as basophils has been difficult with flow cytometry due to their relative scarcity (15, 16) However, wide-spread adoption of this flow cytometry-based test has been limited by the intensive nature of analysis and interpretation required for flow cytometry data by expert operators.
We therefore wrote a data-driven algorithm for flow cytometry data analysis of basophil activation testing (BAT) with the following objectives: transparency and reproducibility, efficiency, quality control, and multiple aggregated statistics of basophil activation. As a proof-of-principle, we applied this script to the analysis of BAT in patients undergoing peanut oral immunotherapy for the treatment of peanut allergy.
Methods
Basophil Activation Assay
Peripheral blood was obtained at multiple time points from pediatric peanut allergic subjects (aged 7–12, positive specific IgE, skin prick testing, and history of reaction, n=30) who underwent a single-center peanut oral immunotherapy trial with peanut antigen (Golden peanut flour)(NCT01324401).
Using the FlowCAST® assay (BÜHLMANN Laboratories, Schönenbuch, Switzerland) and following the manufacturer’s protocol, 0.05 mL of peripheral blood was incubated with peanut allergens: Arah1, Arah2, Arah6, whole peanut extracts (0503 or Golden peanut flour) for 15 minutes at 37°C along with anti-CCR3 PE and -CD63 FOTC antibodies. Control conditions included a medium-only negative control, a positive control involving the crosslinking of the high-affinity Fc epsilon receptor (anti-FcERI), and a positive control independent of FcERI signaling, involving stimulation with N-Formylmethionyl-leucyl-phenylalanine (fMLP). Flow cytometry data was collected on a BD LSRII or BD Fortessa as FCS data files (MiCyte Appendix).
Computational analysis: Autogating
For development of this algorithm, we created the algorithm based on a subset of the BAT experiments from clinical trials, validated it with the standard manual analysis of the same data, and then applied the algorithm to a larger dataset (Figure 1A).
Figure 1.
Analysis of basophil activation tests (BAT).
A. Flow cytometry of antigen-stimulated whole blood is analyzed using data-driven analysis. B. Basophils are identified as SSClo, CCR3hi cells, with activated basophils as CD63hi using manual verses autogating analysis.
Using Bioconductor tools in R (flowCore and associated packages)(17–24), we developed a data-driven, automated gating strategy in which basophils were defined as bivariate normally distributed SSCloCCR3+ cells after removal of SSChi and CD63−CCR3lo populations using clustering approaches (kmeans and norm2Filter). Comparable basophil populations were achieved, as shown above in the same sample. Gate visualization on plots of all stimulation conditions provided additional quality control (Figure 1B, supplemental figures) (https://github.com/saritaupatil/AutoBAT).
Similar to the manual gating, two gating strategies to determine CD63 upregulation were employed. Gating on resting basophils determines CD63 upregulation by setting the cutoff at the 97.5 percentile of CD63 expression of unstimulated basophils. Gating on activated basophils determines CD63 upregulation using k-means clustering to distinguish between bimodal CD63 populations in anti-FcERI treated basophils. The estimated time for implementation of this analysis is about 5 minutes of total screen time by a skilled operator.
Manual analysis
For comparison with the programmatic approach, a subset (n=52) of the experiments were manually gated using FlowJo (Tree Star, USA) software. On manual gating using FlowJo templates, basophils were identified as single cells on FSC-H and FSC-A followed by gating on the SSCloCCR3+ population. These were adjusted as needed after visual inspection of each sample in the experiment (Figure 1B). CD63 upregulation was defined using 2 methods: either on resting basophils, where the cutoff is set at the 97.5 percentile of CD63 expression of unstimulated basophils, or on activated basophils where the cutoff is set at the midpoint of the bimodal populations of the CD63 expression in anti-FcERI stimulated basophils. The estimated time for analysis of each experiment was about 5 minutes. We also used manual gating for data from experiments with limited quality.
Statistical analysis
Statistical comparison of the percentage of CD63hi basophils derived from similar manual and automated gating strategies was analyzed using the Spearman correlation coefficient. Dose response curves were fitted using the R package drc(25) assuming a four-parameter log-logistic dose-response model and the effective dose ED50 was derived. The default, non-robust least square unconstraint estimation fitting procedure (“mean”) was not always the optimal solution, so we also fitted the same dose-response using robust methods(26): median estimation (“median”), least median of squares (“lms”). Because the CD63hi basophils were bounded (they cannot be less than 0 or larger than 100) we also extended the mean fitting procedure by a “constrain” estimation(26) restricting the lower and upper bounds (so-called box constraints) to those allowed by the data. For each of these four estimating procedures we selected the dose-response curve with the smallest AIC (Akaike’s Information Criterion).
Results
Autogating is reproducible and comparable to manual gating
During development of the autogating approach, we first tested a subset of the data, 48 experiments (16% of the total of 294 experiments) with both conventional gating and programmatic approach. CD63 upregulation using conventional gating and the programmatic approach are linearly correlated across all stimulation conditions.
Gating based on resting basophils correlated to respective manual gating more closely than gating on activated basophils (R=0.96 vs. R=0.84, p<0.001) (Figure 2). The correlation of automated and manual analysis significantly depends the activation status (p=0.025) and number of basophils per analysis (p<0.0001). This subgroup analysis highlights again the need for robust BAT data acquisition, which increases the probability of success with automated gating.
Figure 2.
Comparison of basophil gating strategies.
A. The threshold for identification of CD63hi basophils was set either using resting or anti-FcERI activated basophils. B. Comparison with manually gated basophils found that gating based on resting basophils were more reliable (p=0.001). C. Using this strategy, dose-response curves in a single patient over time are reliably identified and are similar across both the autogating (solid line) and manual gates (dashed line).
Analysis of residuals, defined as the difference between CD63 upregulation in the conventional and the programmatic approach, found that only 2.0% of the comparisons fell outside the standard error. Incongruence between manual and automated gating is primarily due to poor basophil identification by the autogating algorithm in one of the experiments used to test the accuracy of the gating algorithm.
Inter-operator variability using conventional manual gating based on resting basophil CD63 fluorescence was performed on 250 samples, revealing a correlation of 0.98.
Data for analysis was collected over 4 years, by multiple operators, and 2 flow cytometers, of which one had a change in laser configuration.
Autogating improves efficiency
In 269 experiments, with a total of 7,166 individual basophil activation samples from 26 subjects, we successfully implemented the above algorithm. We estimate this represents a net saving of 1340 minutes of labor by a skilled operator.
Of the 294 experiments, 25 (8.5%) failed automated gating required more permissive manual gating. Technical errors including experimental error and errors during flow acquisition, as well as poor sample quality contributed to failure of autogating.
Autogating facilitates transparency
Longitudinal analysis of automated basophil flow cytometric characteristics captures aberrations and outliers, which in conjunction with visual quality control, identifies experiments for which automated gating strategy is less robust, likely due to poor quality data. Our strategy also incorporates graphical output for each experiment for visual quality control analysis as desired (Appendix B).
By aggregating data across all of the experimental conditions, we used longitudinal measures of CCR3 MFI and basophil enumeration to identify experiments with limited quality, due to low or absent fluorescence signal or high parameter variance, likely reflecting technical problems with sample acquisition or preparation or low numbers of basophil events reflecting biological variation. Of the 264 experiments, 16 (5.9%) had identified basophils with abnormally low CCR3 expression and 16 experiments (5.9%) had insufficient basophils. All together, 29 experiments (10.8%) were identified as having at least one stimulation condition, with either low CCR3 expression or insufficient basophil numbers, that did not meet our quality threshold (Figure 3). The workflow using this algorithm can also identify experiments which fail on quality control measures and which may require subsequent more permissive or manual gating.
Figure 3.
Quality control measures in BAT.
Longitudinal CCR3MFI and basophil counts in experiments throughout the protocol identifies outlier basophil activation tes ts. All together, 29 experiments (10.8%) were identified as having at least one stimulation condition that did not meet our quality threshold.
Autogating provides additional statistical measures of basophil activation
Various measures of basophil activation, including ED50 (CDsens), area-under-the-curve (AUC), and maximum CD63pos (CDmax) (9, 16, 27) can be calculated using our tools (Figure 4).
Figure 4.
Dose-response measures in BAT.
A. Our analytical script supports quantification of additional measures of the dose-response curve in BAT. The CDmax (dashed line), or maximal stimulation, of the dose response curve can be calculated, as can the ED50 (solid line), which is the antigen concentration at which one-half of the CDmax is achieved. The area-under-the-curve which can be an aggregate measure of the dose-response curve can also be calculated using the trapezoidal approach (grey shaded area). B. Analysis of the basophil dose-responses to peanut antigens using different model fitting for the calculation of AUC and ED50 shown in a single patient before and after therapy. The AIC describing the fit for each dose-response model is shown in the upper left corner of each graph, and the model with the smallest AIC was selected.
In the current BAT database, the AUC was calculated for 99.18% and an ED50 was calculated for 85.02% of the 1094 BAT dose-response curves.
Discussion
In this paper, we adopted the R Bioconductor package, flowCore to implement a kmeans clustering-based gating strategy to analyze 269 basophil activation tests from a clinical trial. We found this strategy to be time-efficient and highly concordant with more conventional, manual gating. The application of a programmatic approach to the BAT provides a high-throughput, reproducible, unbiased approach to flow cytometry data analysis.
By facilitating analysis of entire experiments or ‘flow sets’, programmatic approaches facilitate quality control and easy monitoring of flow cytometry parameters and assay variability over time, which is particularly useful in clinical and diagnostic applications, including multicenter studies. In our dataset, using CCR3 MFI clustering, we were able to readily identify experimental outliers which were the result of poor quality data, whether technical/experimental error or error during data collection. These outliers can be further analyzed using automated analysis, by changing and adapting the algorithm. As this particular basophil assay relies on side scatter as an important flow cytometry parameter, it may have been less robust to these types of technical error which could be a consideration for future studies.
The notable limitation of this analytical approach is that the data-driven programmatic analysis of basophil activation testing is dependent on the acquisition of high quality flow cytometry data, including the use of a well-designed antibody staining panel, setting of appropriate PMTs, lack of technical error, and reasonable basophil counts. In our analysis, the number of basophils acquired within each condition significantly influence the accuracy of the autogating algorithm. Using repeated sampling, basophil degranulation measured by CD63 varies <10% at an N of 100 and <5% at an N of 400. Dose-response measures, such as AUC, vary < 10% at an N of 500. Further validation of this tool would benefit from testing a wide range of antigens, instrumentation, and centers. The use of this algorithm, for example, in multi-center trials would rely on the implementation of flow cytometry parameters that allow for accurate discrimination of different cellular populations, and some real-time review of quality control parameters for the assays, such as a minimal number of basophil events acquired and fluorescence measurements falling within a pre-specified range.
This particular algorithm depends heavily on control tubes, with medium-only and anti-FcERI stimulation to set the parameters for identifying basophil degranulation. During the development of this approach, we observed that setting a cutoff to resolve activated versus resting basophils was more reliable when using the unimodal distribution of the resting basophils compared to the bi-modal distribution of the anti-FcERI activated basophils. Though gating on resting basophils will tend to derive a lower cutoff and may be more sensitive to ‘drift’ in absolute fluorescence of the non-degranulated basophils, gating on activated anti-FcERI requires bi-modal upregulation of CD63, if the basophils undergo piecemeal degranulation, a CD63-intermediate population complicates the identification of a reproducible cut off Furthermore, the fact that the number of anti-FcERI activated basopihls were consistently lower, made the midpoint determination unreliable in a small, but significant number of experiments. We speculate that the lower basophil event numbers in the anti-FcERI activation condition may be partly an artifact of them having increased side scatter characteristics and that together with simply acquiring more total events, the use of more than a single fluorescence channel, might overcome this limitation. Nevertheless, the use of a cutoff derived from the resting basophils correlated well with conventional gating. The presence of non-responders, which can be seen in 10% of the population, is a known limitation of BAT, and the use of resting basophils to derive the cutoff for activation also has the advantage of reliable application in basophil activation non-responders.
In our data, we recognized several sources of variability. Experimental variability in sample preparation and technique had a strong adverse impact on data quality and the study-wide data analysis and QC that is facilitated by an automated gating workflow helped quickly identify this (Figure 3A, low CCR3 MFI events). Technical variation from instrument variation or parameter settings during data acquisition was another key source of variability, and these pieces of flow file metadata can be incorporated into a quality control workflow. Longitudinal studies with repeated measures can be effective in measuring this variability. These types of variability can be mitigated by training, delineating parameters for sample preparation and flow cytometry data acquisition, and utilization of frequent quality control measures during the clinical trial or period of experimentation. Other sources of variability for this assay, including the stimulants and reagent (i.e., antibody) variability was minimized in this flow set by using a pre-standardized and validated basophil activation kit. Finally, significant inter-subject variability in human subjects exists, both at baseline and with therapeutic intervention, as in this dataset, and should be captured for further analysis.
Integrating the derivation of summary statistics for the dose-response curve is another important component of this program. Previous studies have correlated various aspects of basophil activation, including the ED50 (CD-sens) and maximal response (CDmax) with distinct clinical outcomes(6, 7, 16, 27). We have also included the AUC as a summary statistic of the dose response, which can be particularly helpful when dose-responses do not follow the expected sigmoidal shape and curve fitting fails(9, 29). In this data set – derived from a study of peanut oral immunotherapy – we found that accurate ED50 derivation from dose-response curves was computationally challenging when dose-response curves had very low variability, which occurs with strong basophil suppression during therapy. In an attempt to overcome the issue of atypical dose-response curves, we evaluated multiple models, including a least square unconstrained estimation fit, median estimation fit, and least median of squares fit to derive the most robust ED50. Despite this, calculating an ED50 is less computationally tractable than the maximal response or AUC as a measure of BAT.
As flow cytometry data files can be sizable, the computational burden of high-throughput data analysis of a large experiment will be best run on a high performance server, now readily available through cloud computing. Implementation requires basic knowledge of R and Bioconductor.
On the other hand, this tool is highly proficient at conducting high throughput analysis, can readily integrate clinical phenotypic data, allows for reproducible and easily shared data analysis between multiple groups. This approach can be further expanded to include high-dimensional, multi-color flow cytometry BAT experiments, which have the additional advantage of utilizing more sophisticated clustering algorithms to improve both the success rate of the automated gating approach as well as the accuracy (30).
In conclusion, this data-driven automated gating algorithm for basophil activation testing provides a data analysis platform to promote transparency, reproducibility, quality controls, and data sharing in clinical research using basophil activation testing.
Supplementary Material
Acknowledgments
Thank you to Dr. Ravi Mylvaganam, Director of the Flow Cytometry Core at the Charlestown Navy Yard, Massachusetts General Hospital, for his intellectual contribution to the collection of clinical flow cytometry data. Thank you to Drs. Bert Ruiter and Yamini Virkud for their review of this manuscript.
We would like to thank clinical research coordinators for their efforts in sample collection: Dr. Stephanie Kubala, Alisa Brennan, Alex Alejos, Lauren Tracy, Caroline Southwick, and Alanna Hickey. We would like to thank our patients and their families for their dedication and commitment to clinical research.
Cytometric findings reported here were performed in the MGH Department of Pathology Flow and Image Cytometry Research Core, which obtained support from the NIH Shared Instrumentation program with grants 1S10OD012027-01A1, 1S10OD016372-01, 1S10RR020936-01, and 1S10RR023440-01A1.
We would like to thank Demarest Lloyd, Jr. Foundation for their support of the peanut oral immunotherapy trial.
Appendix
Appendix A
Basophil sample number derivation: Using repeated sampling of SSCloCCR3+ basophils from 22 experiments (~10% of the total experiment number), we randomly selected different sample numbers of basophils, ranging from 50 to 1000) 1000 times and calculated the percentage of CD63hi basophils (A) and the AUC of the antigen stimulation dose response curve (B). Variation from the measured value decreases with increased basophil sample numbers.
Appendix B
Graphical output from the data-driven BAT analysis.
Footnotes
Conflict of Interest Disclosure
Experimental support and reagents provided by BÜHLMANN Laboratories.
Bibliography
- 1.Burks AW, Jones SM, Wood RA, Fleischer DM, Sicherer SH, Lindblad RW, Stablein D, Henning AK, Vickery BP, Liu AH, Scurlock AM, Shreffler WG, Plaut M, Sampson HA, Consortium of Food Allergy R Oral immunotherapy for treatment of egg allergy in children. The New England journal of medicine. 2012;367:233–43. doi: 10.1056/NEJMoa1200435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Thyagarajan A, Jones SM, Calatroni A, Pons L, Kulis M, Woo CS, Kamalakannan M, Vickery BP, Scurlock AM, Wesley Burks A, Shreffler WG. Evidence of pathway-specific basophil anergy induced by peanut oral immunotherapy in peanut-allergic children. Clin Exp Allergy. 2012;42:1197–205. doi: 10.1111/j.1365-2222.2012.04028.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Finak G, Langweiler M, Jaimes M, Malek M, Taghiyar J, Korin Y, Raddassi K, Devine L, Obermoser G, Pekalski ML, Pontikos N, Diaz A, Heck S, Villanova F, Terrazzini N, Kern F, Qian Y, Stanton R, Wang K, Brandes A, Ramey J, Aghaeepour N, Mosmann T, Scheuermann RH, Reed E, Palucka K, Pascual V, Blomberg BB, Nestle F, Nussenblatt RB, Brinkman RR, Gottardo R, Maecker H, McCoy JP. Standardizing Flow Cytometry Immunophenotyping Analysis from the Human ImmunoPhenotyping Consortium. Scientific reports. 2016;6:20686. doi: 10.1038/srep20686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lee JA, Spidlen J, Boyce K, Cai J, Crosbie N, Dalphin M, Furlong J, Gasparetto M, Goldberg M, Goralczyk EM, Hyun B, Jansen K, Kollmann T, Kong M, Leif R, McWeeney S, Moloshok TD, Moore W, Nolan G, Nolan J, Nikolich-Zugich J, Parrish D, Purcell B, Qian Y, Selvaraj B, Smith C, Tchuvatkina O, Wertheimer A, Wilkinson P, Wilson C, Wood J, Zigon R, International Society for Advancement of Cytometry Data Standards Task F. Scheuermann RH, Brinkman RR. MIFlowCyt: the minimum information about a Flow Cytometry Experiment. Cytometry Part A : the journal of the International Society for Analytical Cytology. 2008;73:926–30. doi: 10.1002/cyto.a.20623. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.O’Neill K, Brinkman RR. Publishing code is essential for reproducible flow cytometry bioinformatics. Cytometry Part A : the journal of the International Society for Analytical Cytology. 2016;89:10–1. doi: 10.1002/cyto.a.22805. [DOI] [PubMed] [Google Scholar]
- 6.Santos AF, Du Toit G, Douiri A, Radulovic S, Stephens A, Turcanu V, Lack G. Distinct parameters of the basophil activation test reflect the severity and threshold of allergic reactions to peanut. J Allergy Clin Immunol. 2015;135:179–86. doi: 10.1016/j.jaci.2014.09.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Santos AF, Lack G. Basophil activation test: food challenge in a test tube or specialist research tool? Clinical and translational allergy. 2016;6:10. doi: 10.1186/s13601-016-0098-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Patil SU, Shreffler WG. BATting above average: basophil activation testing for peanut allergy. J Allergy Clin Immunol. 2014;134:653–4. doi: 10.1016/j.jaci.2014.07.040. [DOI] [PubMed] [Google Scholar]
- 9.Patil SU, Shreffler WG. Immunology in the Clinic Review Series; focus on allergies: basophils as biomarkers for assessing immune modulation. Clinical and experimental immunology. 2012;167:59–66. doi: 10.1111/j.1365-2249.2011.04503.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zidarn M, Kosnik M, Silar M, Bajrovic N, Korosec P. Sustained effect of grass pollen subcutaneous immunotherapy on suppression of allergen-specific basophil response; a real-life, nonrandomized controlled study. Allergy. 2015;70:547–55. doi: 10.1111/all.12581. [DOI] [PubMed] [Google Scholar]
- 11.Burton OT, Logsdon SL, Zhou JS, Medina-Tamayo J, Abdel-Gadir A, Noval Rivas M, Koleoglou KJ, Chatila TA, Schneider LC, Rachid R, Umetsu DT, Oettgen HC. Oral immunotherapy induces IgG antibodies that act through FcgammaRIIb to suppress IgE-mediated hypersensitivity. J Allergy Clin Immunol. 2014;134:1310–17 e6. doi: 10.1016/j.jaci.2014.05.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Santos AF, James LK, Bahnson HT, Shamji MH, Couto-Francisco NC, Islam S, Houghton S, Clark AT, Stephens A, Turcanu V, Durham SR, Gould HJ, Lack G. IgG4 inhibits peanut-induced basophil and mast cell activation in peanut-tolerant children sensitized to peanut major allergens. J Allergy Clin Immunol. 2015;135:1249–56. doi: 10.1016/j.jaci.2015.01.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Syed A, Garcia MA, Lyu SC, Bucayu R, Kohli A, Ishida S, Berglund JP, Tsai M, Maecker H, O’Riordan G, Galli SJ, Nadeau KC. Peanut oral immunotherapy results in increased antigen-induced regulatory T-cell function and hypomethylation of forkhead box protein 3 (FOXP3) J Allergy Clin Immunol. 2014;133:500–10. doi: 10.1016/j.jaci.2013.12.1037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Thyagarajan A, Varshney P, Jones SM, Sicherer S, Wood R, Vickery BP, Sampson H, Burks AW. Peanut oral immunotherapy is not ready for clinical use. J Allergy Clin Immunol. 2010;126:31–2. doi: 10.1016/j.jaci.2010.05.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Eberlein B, Hann R, Eyerich S, Pennino D, Ring J, Schmidt-Weber CB, Buters J. Optimizing of the basophil activation test: Comparison of different basophil identification markers. Cytometry Part B, Clinical cytometry. 2015;88:183–9. doi: 10.1002/cyto.b.21203. [DOI] [PubMed] [Google Scholar]
- 16.Hoffmann HJ, Santos AF, Mayorga C, Nopp A, Eberlein B, Ferrer M, Rouzaire P, Ebo DG, Sabato V, Sanz ML, Pecaric-Petkovic T, Patil SU, Hausmann OV, Shreffler WG, Korosec P, Knol EF. The clinical utility of basophil activation testing in diagnosis and monitoring of allergic disease. Allergy. 2015;70:1393–405. doi: 10.1111/all.12698. [DOI] [PubMed] [Google Scholar]
- 17.Ellis B, H P, Hahne F, Le Meur N, Gopalakrishnan N, Spidlen J, Jiang M. flowCore: flowCore: Basic structures for flow cytometry data. 2016 [Google Scholar]
- 18.Ellis B, G R, Hahne F, Le Meur N, Sarkar D. flowViz: Visualization for flow cytometry [Google Scholar]
- 19.Hahne F, G N, Khodabakhshi AH, Wong C, Lee K. flowStats: Statistical methods for the analysis of flow cytometry data. 2016 [Google Scholar]
- 20.Huber W, Carey VJ, Gentleman R, Anders S, Carlson M, Carvalho BS, Bravo HC, Davis S, Gatto L, Girke T, Gottardo R, Hahne F, Hansen KD, Irizarry RA, Lawrence M, Love MI, MacDonald J, Obenchain V, Oles AK, Pages H, Reyes A, Shannon P, Smyth GK, Tenenbaum D, Waldron L, Morgan M. Orchestrating high-throughput genomic analysis with Bioconductor. Nature methods. 2015;12:115–21. doi: 10.1038/nmeth.3252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Spidlen J, G N, Hahne F, Ellis B, Gentleman R, Dalphin M, Le Meur N, Jiang PaW B. flowUtils: Utilities for flow cytometry. 2016 [Google Scholar]
- 22.R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2015. [Google Scholar]
- 23.Gentleman R, H F, Kettman J, Le Meur N, Gopalakrishnan N. flowQ: Quality control for flow cytometry. 2016 [Google Scholar]
- 24.Team RC. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2015. [Google Scholar]
- 25.Ritz Ca, S JC. Bioassay analysis using R. Journal of Statistical Software. 2005;12:1–22. [Google Scholar]
- 26.Ritz C, Baty F, Streibig JC, Gerhard D. Dose-Response Analysis Using R. PloS one. 2015;10:e0146021. doi: 10.1371/journal.pone.0146021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Nopp A, Johansson SG, Ankerst J, Bylin G, Cardell LO, Gronneberg R, Irander K, Palmqvist M, Oman H. Basophil allergen threshold sensitivity: a useful approach to anti-IgE treatment efficacy evaluation. Allergy. 2006;61:298–302. doi: 10.1111/j.1398-9995.2006.00987.x. [DOI] [PubMed] [Google Scholar]
- 28.MacGlashan D., Jr Expression of CD203c and CD63 in human basophils: relationship to differential regulation of piecemeal and anaphylactic degranulation processes. Clin Exp Allergy. 2010;40:1365–77. doi: 10.1111/j.1365-2222.2010.03572.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.MacGlashan DW., Jr Basophil activation testing. J Allergy Clin Immunol. 2013;132:777–87. doi: 10.1016/j.jaci.2013.06.038. [DOI] [PubMed] [Google Scholar]
- 30.Chester C, Maecker HT. Algorithmic Tools for Mining High-Dimensional Cytometry Data. J Immunol. 2015;195:773–9. doi: 10.4049/jimmunol.1500633. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.