labeled antibodies used to stain the sample. Compensation matrices derived from mix-and-match single-stained bead listmode files are computed and applied during data analysis. When tandem dyes are UJ1d('r~tdnding Clinical Flow Cytometry 20'; used. it is 1101 unusual to create four separate cOlllrensatio11 matrices for a panel consisting of four tubes. Data-presentation aids that allow on-axis evcnh to be visualized (e.g.. Baseline Offset from Beckman Coulter and Log Bias from Verity Software House) must never be used during compensation, because even grossly overcompensated events appear to be perfectly compensated. With the exception of DNA, ploicly measurements, ancl a few othcr spcci fic appl ications, most fluorescence clata are conventionally displayed on a logarithmic scale. Compared to a linear scale, a log scale emphasizes (spreads out) lower values and compresses higher ones. It also normalizes the distribution of fluorescence intensity within a given cell population, and is a very useful way of visualizing data that is distributed over a wide dynamic range. Zero ancl negative numbers are undefined on a log scale, and this often creates a problem causing data to pile up on the axes. A new family of data display scales (biexponential, hyperbolic sine function, HyperLog [59]) offers the best of linear and log displays. These scales are approximately linear on the low end of the scale and become logarithmic at higher values. Like a linear scale, 0 and negative values are defi ned. This is very useful for evaluating compensation, since the mean FIBComp for a sample single-stained with A should be O. Because measurements have variance (spread), a symmetrical distribution of the data around the mean 01'0 means that there will be both positive and negative values. This is easily visual ized ina HyperLog plot (Figure 6.12), but missed entirely ina conventional log display where 0 and negative values are undefined, and events with fluorescence less than or equal to the lowest value defined in the scale are crowded together on the axis. It will take time before this type of display is universally accepted and it is not without problems (determining how much graphic space to devote to negative values Logarithmic scale Hyperlog scale .': 0 0 0 0 w W 'l" 'l" (') .' (') 0 0 0 0 CD45 APC-Cy7 CD45 APC-Cy7 FIGURE 6.12 Use of the HyperLog scale to evaluate spectral compensation. In this example, the same data are displayed on a conventional logarithmic scale (left panel) and on a HyperLog scale (right panel). The origin of the quadrants on the right panel indicates fluorescence intensities of zero. Zero and negative values are undefined on the logarithmic scale, and all such values pile up on the axes. When data are properly compensated, negative populations are symmetrically distributed abollt the origin. b 206 Handbook of Human Immunology, Second Edition is dirficult because the range or negative values can be quite variable). However, it solves rar more problems than it creates, and is likely to become the standard. One more word on compensation standards: When using tandem dyes we strongly recommend against using cells stained with a bright antibody (e.g., CD45 PE-Cy7, CD8 APC-Cy7) as a compensation standard for other antibodies conjugated with the same tandem dye. Although this is recommended by some cytometer manufacturers, it makes the assumption that both antibodies require the same amount of compensation. This is asking for trouble. 6.12 QUALITY CONTROL FOR ABSOLUTE COUNTS The single-platform absolute count method using the lyse no wash method is widely used for determining absolute CD34 or T-cell counts (i.e., quantifying the concentration in cells/microliters). The principle is simple. According to one popular method, a known and carefully measured volume ofsample is deposited in a tube and stained by the addition of directly labeled monoclonal antibodies. After staining, the red blood cells are lysed by the addition of a large excess of ammonium chloride-based lysing solution. Calibration beads of known concentration, of volume equal to the sample volume and measured with equal care, is then added. After gentle mixing, a viability dye is added and the sample is