An example:
Let's say you're viewing FSC, p-STAT1-Ax448, and p-STAT5-Ax647 in the rows of one heatmap plot.  In the columns are timepoints.  You want to compare within each row to the first column and you pick "Medians" as the statistic.

1) Raw (no equation)
For each row you will see the raw, untransformed values on a scale ranging from min to max (e.g. -200 to 262,144).  Values might be 10 and 1,000.  The color change is linear, so the heatmap will look the same as if you did Fold.  All three channels (the rows of the heatmap) will use raw values.

2) Fold
For each row you will see the fold of the raw value vs. the control (x/control).  Values might be 1 and 100.  The color change is linear so the change from 2 to 10 will be given as much weight as the change from 200 to 2000 (which is incorrect for e.g. LSRII data but correct for Calibur data).  The cofactor is not used.  All three channels (the rows of the heatmap) will use fold.

3) log10 ratio
For each row you will see the log10 ratio (log10(x/control)).  Values might be 0 and 2.  The color change is non-linear (it follows the log10 equation).  The change from 1 to 10 will be given as much weight as the change from 100 to 1000 (which is incorrect for e.g. LSRII data but correct for Calibur data).  The cofactor is not used.  All three channels (the rows of the heatmap) will use log10.

4) Arcsinh ratio
For each row you will see the arcsinh ratio (arcsinh(x/control)).  Values might be 0 and 3.5.  The color change is non-linear (it follows the arcsinh equation), which will de-emphasize changes near the axis in proportion to the size of the cofactor for that channel.  If the cofactor is properly set, this is the correct way to display and compare LSR II data.  If the cofactor for that channel is 1, the heatmap will look the same as if you did log10 ratio.  All three channels will (the rows of the heatmap) use arcsinh.  Each row uses its own cofactor.

5) Transformed ratio
For each row you will see a comparison using the scale of the display.  In our example this would likely be linear for the first row (FSC) and arcsinh ratio for the second two rows (p-STAT1-Ax448 and p-STAT5-Ax647).  Each row uses its own cofactor.