Interpreting Results

QTube presents you with a color-coded table of values for the metrics for each tube. Let's look at how it gets from your FCS files to the final result.

Example:  samp1

The example we'll use is taken from the Example Data. The dotplots for CD45 vs Side Scatter for Tubes 1 and 6 for the "samp1" sample are shown referenced to contours from the aggregate of the first 6 tubes in the panel (tubes 7-10 were permealized for intracellular labels, so we won't consider them for now):

Notice that Tube 1 (which is closely representative of all but Tube 6) lines up nicely with the contours, while Tube 6 shows a subtle shift towards the origin. This is the type of defect we hope QTube will notice.  Depending on your particular analysis, such a shift may or may not be problematic.  You can tailor QTube's sensitivity using the controls in the Preferences [    ] panel.  

Low Resolution

We've run QTube on this example (tubes 1-6), at a Low resolution and using Side Scatter and CD45. The results are shown as follows:

Note that Tube 6 looks a bit worse than the others - as expected, knowing in advance what the dotplots look like. Let's examine how this result is arrived at. Take a look at the Max Metric first. Remember that this metric is the maximum absolute value of the base-2 logarithm of the fingprint. Let's plot the fingerprints on a base-2 log y axis:

The Tube 6 fingerprint is shown in blue, the others in black. Note that the Tube 6 fingerprint deviates significantly from 0. The other fingerprints are all hovering pretty closely around 0 (which is the log of 1.0, so the ratio of the densities in these bins is close to the expected density for the aggregate) and none of them gets outside of the green.

High Resolution

Now, we've used the same six tubes, this time using 'High' resolution. Here's what the results look like:

Note that now the Max Metric looks a little worse. Why is that? To answer this question we'll again look under the hood at the fingerprints themselves:

Notice that the fingerprint is "denser".  That's because the bins into which the data have been divided are smaller, so it takes more of them to cover the space.

The Log2 fingerprint for Tube 6 is again drawn in blue, the others in black. Notice that the Tube 6 fingerprint wiggles around a bit more - we're measuring density in smaller regions, so local fluctations aren't averaged out as much as at lower resolution. Also note that where the Tube 6 fingerprint is very high, the others are a bit low (and vice versa). This is because the higher density that Tube 6 contributes to the aggregate in that subregion has "pulled" bins towards it, making them smaller, and leaving less volume for events in the other tubes. Therefore, the other tubes extend slightly into the yellow region. Another way of looking at this is that the average value of a fingerprint element across the tubes will be 1.0, so if one of them is anomalously high, the others will be slightly lower such that their average remains 1.0.

The Standard Deviation Metric

If we collect all of the fingerprint values together for each tube, we can calculate their standard deviation.  That's what this metric is all about.  Looking at the boxplot below we can see that the variance for Tube 6 is much higher than the other tubes.   Given our choice for the colors ranges, this lands Tube 6 in the yellow.  This metric is useful for looking across the entire fingerprint and not zeroing in on a single anomalous bin.  

Some final thoughts

It's important that you develop your own intuition using your data in QTube.  Experiment around with data you know to be good, and other data that you think have some inconsistencies such as the ones we've illustrated.  By going back and forth between QTube results and your favorite way of looking at your data you will begin to understand where you should set resolution and color values for your application.