Today's topic is benchmarking. This is probably particularly relevant for anyone who manages a shared resource lab with an Aurora.
On the Aurora, we can record the spectra of fluorophores and save these for future use in the reference library of SpectroFlo. These stored signatures can be used for unmixing in place of freshly acquired controls, but the accuracy of these spectra declines with time despite the software attempts to update them in line with the daily QC. So, don't rely on these too much for unmixing.
What we can do is use these as a visual guide to assist us in correctly extracting the fluorophore signature from our freshly acquired single-stained controls. This is better than using the webtool spectrum viewers for two reasons: 1) the signature is correct for your machine and your fluorophore:antibody conjugates, and 2) the signature is visible to all users while doing unmixing (provided the benchmarking is done in the Admin account).
To benchmark controls, we first need to store them as reference controls. Here's the process for that on the Aurora. In this example, I'll be storing GFP from iCas9-GFP-expressing mouse cells. The Cas9 is expressed in T cells, so both positive and negative cells will be present in the sample, which is important for defining the spectrum later.
Navigate to the QC and Setup menu.
Select the Reference Controls tab.
Select the fluorophores you want to store as Reference Controls.
Define what your controls are (cells/beads, marker).
Adjust the instrument gains (FSC, SSC, threshold) as appropriate for your sample.
Acquire and record the sample(s).
Adjust the gates to clean up the spectrum. Save the reference spectrum.
See this post on unmixing for more details about why the gates are important.
Finally, select the control(s) you wish to benchmark in the QC & Setup tab under Reference Controls/Benchmark.
Once benchmarked in an Admin account, these reference spectra will appear as a red trace overlay in the unmixing for all users. If the user’s control doesn’t match very well, it will be flagged and will hopefully trigger some troubleshooting. This can help pick up errors such as tandem breakdown, putting the wrong antibody (or multiple antibodies) in the control, autofluorescence noise or selecting the wrong fluorophore in the software (e.g., FITC rather than GFP).
Here's an example of what you should see when performing unmixing on a machine with benchmarked controls:
As you can see, we get the red trace and a Similarity Index for the GFP, which I've benchmarked (in this example, the other fluorophores are not benchmarked). If the Similarity Index falls below a certain threshold, a warning icon will appear. This is a sign to go back and check your single color control. A control that is a bit off (for example, containing additional small peaks from autofluorescence) can be fixed by adjusting the gates, as detailed here. Large differences indicate improper control preparation, fluorophore breakdown, mix-ups or other serious issues.
Hope this helps!
Comments