Compensation is not dependent on intensity, BUT the accuracy at which you determine compensation is dependent on intensity. The brighter the control the more accurately compensation is determined.
Classification of Kerogen, Perspective on palynofacies in depositional envi...
Flow Cytometry Compensation Bright vs Dim Controls
1. Compensation Issue:
Bright vs. Dim Controls
Rich Hastings
KUMC Flow Cytometry Core Lab
3901 Rainbow Boulevard
Kansas City, KS 66160
913-588-0627
rhastings@kumc.edu
http://www.kumc.edu/flow/
2. Compensation
Compensation/Spectral Overlap
Most fluorochromes and dyes excited by laser light have long emission curves.
Flow cytometers have filter sets optimized for specific wavelengths of light.
Unfortunately, overlapping emission wavelengths from one fluorochrome may
spillover into the filter of another.
This spillover can be corrected mathematically by a process called compensation.
Once fluorescence compensation is set for a sample, the compensation values
remain applicable for successive dim or bright samples, because compensation
subtracts a percentage of the fluorescence intensity.
3. Compensation
Misconceptions:
A compensation value > 100% is a problem, no it’s not. It is what it is!
Compensation is not dependent on brightness, compensation is dependent on
the emission spectra of a fluorochrome.*
Same experiment on same instrument, does compensation becomes
unnecessary?
1. How long were your samples exposed to light or fixative?
2. How careful were you in the set up of your stained cells and the
instrument? Does your instrument drift?
3. Check your compensation matrices over time and apply different
compensation values to different experiment days. Trust but verify!
Beads and cells cannot be both used to generate compensation values. Sure
they can!
Compensation introduces errors. Not true, but the data may spread .
Slide adapted from Mario Roederer (Current Protocols in Cytometry)
4. Compensation
Compensation is and is not dependent on brightness. Why the
Ambivalence?
Possible to get the proper compensation value using a dim compensation
control.
Compensation is just much more accurate using bright controls.
Digital Compensation is dependent on complex matrix algebra.
This matrix algebra is the simultaneous solution of the equations for the
contributions of the spectral overlaps of each of the colors into every detector.
(BD)
First Error: Photoelectron counting statistics dictate that the variance of a dim
or “negative” signal will be higher than that of a positive signal. (Howard Shapiro)
Second Error: The loss of information when a signal is converted to an integer
is called digital error. (Bruce Bagwell)
15. Compensation
I clearly manipulated this experiment very bright and very dim APC
controls.
Very possible to get accurate compensation with a dim control, just more
likely with a bright control.
If your dim control is your only compensation control for a specific color
and is the only version of that color in your experiment proper
compensation.
Instrument counting errors and the conversion to digital may create errors
in compensation.
Know your data, what should the data look like?
We correct this spillover mathematically using a process called compensation.
* We will talk more about this on the next slide.
Your compensation control must stain at about the same intensity as the cells in your experiment. Staining intensity does not affect compensation, as long as your comp control is brighter than your samples.
Compensation is not dependent on intensity, BUT the accuracy at which you determine compensation is dependent on intensity. The brighter the control the more accurately compensation is determined.
We used to do compensation in hardware and it could not be changed post-aquisition, if we were doing FITC vs PE, it was easy but compensating 4 or 5 colors- too complex.
How smart were these guys? They anticipated digital instruments and they did this with data collected on a Digital Vax transferred to an early 90’s PC.
I do not understand the math involved. Bruce’s slide from ISAC 2004. But I know when my data is under- or over-compensated.
Let’s do an experiment with dim and bright compensation controls.
Unstained control. Our markers will be APC-H7 vs APC.
Dim and bright APC comp controls, Dim control is barely positive. I cheated I made the CD15 very dim.
Here we are looking at an overlay of the negative control (red), blue (dim CD15 APC) and orange (bright CD309 APC) beads. You can clearly see that the dim population is right-shifted compared to the control. The bright beads are 125x brighter than the dim APC beads.
Here is my bright APC-h7 control.
Look at the compensation values, they are almost double for the dim control as compared to the bright control.
Using the bright CD309 APC as my APC comp control, I am able to correctly compensate both my bright and dim APC populations out of the APC-H7 signal.
When I use the dim CD15 APC as my comp control, I correctly compensate my dim CD15 APC compensation control but I grossly over-compensate my bright CD309 APC stained beads.
Do you even need compensation? This is a very clever experiment. PBMC stained with CD8 PE and CD3 FITC, a FMO control with CD3 FITC, then looking at the CD3 FITC and CD8 PE stained control, it’s easy to see where the double positives are and gate them. Obviously, this is a very simplistic experiment, if you add more colors this method quickly gets out of control.