Bioconductor offers packages for R supporting the complete work-flow to analyse large-scale flow cytometry data. We evaluated the usability of this software and compare the variability between manual and automated analysis using a large data set obtained from the analysis of regulatory B- and T-cells from patients with rheumatoid arthritis (RA).
PBMCs from 61 patients suffering from RA were isolated and stained for regulatory T and B cells using two panels containing eight markers each. Additionally, isotype controls were included for two non-lineage markers. Data were acquired using a FACS Aria II flow cytometer.
Manual analysis was performed by two independent evaluators using the FACS Diva software (v6.1.2). For automated analysis, R (v2.15.0) and Bioconductor packages (v2.10) were used. Correlation analyses of the percentages of ten cell populations present in the stained and the non-lineage isotype controls were carried out to determine gating robustness.
Manual analysis took 25 working hours before a table was completed that contained ten variables for subsequent statistical analyses. In contrast, the automated data analysis took 15 hours only. Manual analysis resulted in median correlation coefficients for the percentages of the various cell population in the stained and non-lineage isotype control samples of 0.88 and 0.91 for an unexperienced and an experienced evaluator, respectively. Likewise, the automated analysis resulted in a median correlation coefficient of 0.84 without any significant differences between the three analysis groups. So far, we did not find any correlation between the percentages of CD4+foxp3+CD25+CD127+/-helios+/- cells and the percentages of naïve, memory and transitional and regulatory B cells and their expression of CD80 and CD86.
The freely available statistical software R in combination with Bioconductor's flow cytometry packages is well suited to perform high quality flow cytometry analyses and is comparable with manual analysis in terms of gating precision yet is less time consuming for large-scale studies.