Abstract
““Wait, when was QC last run???” Evaluating MFI drift after morning QC and its impact on unmixing.”
David Rach1, Mikayla Trainor2, Natarajan Ayithan2, Xiaoxuan Fan2
1 Molecular Microbiology and Immunology Graduate Program, University of Maryland School of Medicine, Baltimore, USA 2 Flow Cytometry Shared Resource, University of Maryland Greenebaum Comprehensive Cancer Center, Baltimore, USA
At our core, quality control (QC) beads are run daily on Cytek Aurora instruments as part of the instrument startup process. Based on the beads initial observed MFI values for each detector, the SpectroFlo software adjusts the gains and laser settings of the instrument to ensure that the after-QC MFI values match lot-specific thresholds. This accounts for instrumental changes, allowing specimens acquired on different days to be comparable, and reducing the frequency of MFI-based batch effects in large spectral flow cytometry panels. We recently implemented a website to monitor daily QC for our instruments at our core https://umgccfcss.github.io/InstrumentQC/, visualizing longitudinal changes in gain, RCV, and bead MFI values before and after daily QC. In the process, we observed various detectors for which MFI values pre-QC were consistently different from the MFI values observed after QC the day before. While MFI was reset to baseline at the following QC, we were curious whether these observed MFI drifts would have already occurred by the evening before, or were a result of the instrument shutdown. Additionally, we wanted to evaluate whether these drifts were sufficient to impact unmixing when samples were acquired in the evening compared to shortly after morning QC.
To evaluate this, we acquired 5000 SpectroFlo QC beads as fcs files, i) before morning QC, ii) after morning QC, iii) before evening QC, and iv) after evening QC. These samples were acquired on a 3, 4, and 5-laser Aurora over a several month period, for both the 2005 and 2006 SpectroFlo QC bead lots. For analysis, acquired FCS files were imported to R, singlet beads gated, and gate placement validated using the flowWorkspace, openCyto and Luciernaga R packages. From the gated events, median MFI and RCV values were calculated for each detector, and voltage/gain metadata for individual .fcs files was retrieved using the Luciernaga package. We then visualized this data in R using various tidyverse packages. We observed that for most detectors, there were limited changes in MFI values between the After Morning QC and Before Evening QC timepoints. However, we noted consistent and significant shifts in MFI for a few detectors by the time of evening QC, notably the YG2, YG3, and R1 detectors. To evaluate whether these observed drifts in MFI would have altered normalized fluorescent signature, we simulated the equivalent day-specific adjustment to reference signatures of over 100 fluorophores, plotting the adjusted signatures against the original reference signature. We observed that the drift in the handful of detectors did not significantly alter the normalized fluorescent spectra of the fluorophore, with exception of a few fluorophores where the detector in question landed on the secondary peak of the spectra. When evaluating the signatures by their cosine value, the differences were within range we would anticipate limited impact on unmixing.
Finally using the same approach, we adjusted raw reference controls and full-stained samples acquired following morning QC and imported into SpectroFlo for unmixing. We did not observe any major changes to the unmixing in medium-sized panels. In summary, for the instruments at our core, the majority of detector MFI values remain stable following morning QC. Additionally, for the few detectors that did consistently shift, based on both simulated and experimental data, the observed changes would have minimal impact for most fluorophores signatures and subsequent unmixing. Whether these observed small changes are enough to affect unmixing in large panels (40+) colors remains an area that merits further investigation.
License
In our commitment to open-science and open-source, all teaching materials are freely offered under a CC-BY-SA license, while all code examples are offered under the AGPL3-0 copyleft license.
