Protocol Online logo
Top : New Forum Archives (2009-): : Flow Cytometry

problems with compensation PE & FITC - (Apr/16/2013 )

Hi, I use three different fluoroforms in my experiment - CD4-PE , human Enolase-FITC and TRX-APC/Cy7 and try to compansate correctly especially PE & FITC. I think I do it correctly - eg stain my population with hE-FITC and then mix it with unstained population ( all done on PBMCs ) but when I am trying to do the compensation it comes to about 75% (FL2:FL1) - way too much?
I use flow cytometer with two lasers (five filters).
Did anyone have similar problem? I already did some troubleshooting and I am about to try again but will welcome any suggestions,thanks



A good way to check if the compensation is accurate is to see if the geometric mean of your unstained population matches up with the stained population's mean in a different channel. Let's use an example to illustrate:

Let's say we are compensating for PE. In our compensation tube, we will have CD4+ cells, and ideally some CD4- cells. If we add CD4-PE antibody to the tube, two populations will appear along that PE axis. However, the two populations should be equally negative in every other channel Remember, we only added CD4-PE to that tube; not any other antibodies.

So, if we plot CD4-PE on the y-axis and another fluorochrome (FITC, in your case) on the x-axis, we should see two populations:

Next, you want to make a quadrant gate to the right of those two populations so that PE+/FITC+ is Q1 (empty), PE+/FITC- is Q2 (has events), PE-/FITC- is Q3 (has events), and PE-/FITC+ is Q4 (empty). Then you need to take the geometric means of your populated quadrants (Q2 and Q3, in this case) and compare them. That step varies a little bit depending on your software, but it's usually found under a statistics menu after selecting an individual quadrant. With a correct compensation, the means should be roughly equal and the populations should appear "lined up" along the x-axis, since they are equally negative for that signal. If the means are very different, you need to adjust the matrix values until they are close. I would aim for a difference of ~50 or less (e.g. - 1256 and 1198). Obviously, the closer the means are, the better your compensation, but as long as you're close it should be fine.

Hope that helps!


thank you, JMoynihan! the excellent explanation? very clear even without pictures of plots! :)