Re: FL3/FL4 compensation on a Calibur: Let's get it right!

From: Mario Roederer (Roederer@drmr.com)
Date: Fri Jul 28 2000 - 18:29:29 EST


Since issues about compensation form the basis of a majority of my
pet cytometric peeves, I found it necessary to mount the soap box yet
again...

First I want to correct something David said which is a very common
misconception:

>Compensation is always more difficult when you have two stains that
>are vastly different in relative intensity.  The more disparity, the more
>compensation needed.

This is NOT correct.  Compensation is INDEPENDENT of intensity.
Compensation is ONLY related to the emission spectrum of the
fluorochromes, not how much light is being emitted.  The whole
principle of compensation is based on the fact that it is independent
of intensity.  Again, I would direct nonbelievers to my web pages on
compensation for a full explanation.

Furthermore, your remaining explanation requires correction and comments:

>Your FL4-%FL3 problem is exacerbated by the fact that it
>isn't JUST a compensation problem.  Compensation is used to remove the
>fluorescence contribution due to spectral overlap among fluors.  For instance,
>you have to compensate your "green"(FITC/FL1) dye from the "orange"
>(PE/FL2)channel because it has an orange "tail". We express this as
>FL2-30%FL1,
>for example.  What we're saying is that an amount of light equal to 30% of the
>light detected in FL1 is showing up in FL2 and we need to ignore that much of
>the contribution to signal in that detector.  The problem you're seeing isn't
>due to to spectral overlap.  You see, the CY5 portion of your tandem dye,
>tricolor, is excited very well by your little red diode laser.  It's "real"
>fluorescence in FL4.  (Imagine trying to compensate PE fluorescence
>out of FL2.
>It doesn't work so well.)    So, the reason your FL4(APC?) signal is being
>ablated is because it is lower than the siganl you are getting from tricolor.
>If you compensate the tricolor signal in FL4, you are knocking your dim FL4
>signal off scale. You are also correct that there is a time delay calculation
>that has to be performed by the cytometer to ensure that signal collected when
>cells strike the first laser is correlated with signal from excitation by the
>second.   You should run FACSComp and do the time delay calibration every time
>you are using the diode laser.  If you're not using FL4, turn it off.

This is also mostly incorrect.  There is no difference between "real"
and any other kind (including "tails") of fluorescence.  The
fluorescence of Cy5PE in the FL2 channel are from photons emitted
from the PE molecule.  The fluorescence in the FL3 channel are some
of the very "red" photons emitted from the PE molecule, but primarily
those which come from the Cy5 through resonance energy transfer from
the PE molecule.  Finally, the fluorescence in FL4, as you correctly
note, is from photons coming from direct excitation of the Cy5
molecule by the second laser.

(By the way, I have no problem compensating PE fluorescence out of
FL2--in fact, this is the only 100% exact compensation that can be
performed!  Fortunately, our hardware and software interfaces don't
give me a control to perform this perfectly valid, although
admittedly useless, operation.)

However, these are all "real" photons, and all of the measurements
are entirely independent.  Also, because the number of Cy5 photons in
the FL4 channel will be proportional to the amount of Cy5PE, as is
the case for the number of photons in the FL3 channel coming from
energy transfer, you can perfectly well compensate these two channels.

It is necessary to realize that "spectral overlap" refers not only to
emission overlap, but to excitation overlap as well.  Fluorescence is
composed of two distinct spectra:  excitation and emission!

Finally, the time delay calculation is irrelevant in this whole
process....  As long as it was correctly set at the time of
collecting the compensation controls (or bead calibration), then it
will NOT affect the ability to properly compensate the samples.

Finally, realize that you CAN often compensate very large
contributions out of a channel and still see dim fluorescence.
(Indeed, in some of our early 7-color work, we had some compensation
settings as large as 600%!).  The ability to accurately do so has
only to do with the counting statistics--i.e., if you are getting a
lot of photons in each detector, then you can accurately remove the
FL3 contribution from the FL4 detector and still see dim APC
fluorescence.  In real life, this process has some inaccuracy as the
number of photons becomes limiting (not to mention log amp
inaccuracies, which is another story), and therefore the ability to
measure dim FL4 signals is degraded.

>And that's all I have to say about that.

Amen.

mr



This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:55:58 EST