This page is under construction; stay tuned for FlowJo Version 10 with Population Comparison.

In FlowJo, the Population Comparison platform is used to compare single or multiple parameters from a test sample to a single control sample OR a composite of more than one control sample.

  1. One control

The most frequent use case for this platform is to compare each unknown sample to a control.  The control may be an FMO control, a biological (unstimulated), or isotype control.  Standard gating techniques may not work well when samples overlap significantly or there is only a slight shift when comparing an unknown sample to a control sample.  Therefore, this platform will apply several sophisticated algorithms to the data to calculate the difference between two samples.  An overview of the algorithms used is presented here.  Using these algorithms is a more objective approach than manually gating.

  1. Multiple Controls

This platform can also be used to compare each unknown to a composite of several controls.  When comparing multiple samples against each other, it is sometimes not possible (or meaningful) to assign a single sample as the control.  In such a case, the Population Comparison platform can concatenate all of the control samples in order to use the average of all the control samples for comparison. This process mitigates the potential artifact introduced by the selection of a sample as control that might be significantly different than the expected control sample.

Determining the samples to be concatenated is the best approach for an iterative process. One can concatenate all the control samples and compute the distances of each control sample to the average of all of them. Thus, those control samples which are outliers can be removed from the control set. When using only one control, caution is warranted since reduction of the number of samples entered as controls can lead to sampling bias.

Launching and Using the Population Comparison Platform

To start, select the gated or ungated population from your unknown sample.  Go to Tools>Biology>Compare Populations.

Next select the control sample(s) by selecting (will become shaded) and then dragging them to the top middle of the comparison window.  Note that you can select more than one by holding down the shift key.

By default, the statistics for ALL parameters will be computed for every algorithm and then displayed in a tabular format within the interface. Use the button in the top right corner to batch edit the parameters being displayed (if you only want specific parameters in the table).

Check the boxes on the left to choose the Parameters to Display, which will display the graphs for those parameters.

In the workspace, the population comparison ‘node’ will display under the unknown sample.  Dragging this node to another sample (or to the group) will batch process the platform for all samples, while maintaining (locking) the same control(s).

The comparison node can be placed into the table editor for quick batch processing of the statistics.

The comparison node can also be placed in the layout editor for graphical reports and batch processing the graphical reports.

Further Information/Hints

The number of binscan be set manually.  The number of bins that the test and control sample are divided into should be maximized to most easily detect small differences between populations; however, the number of bins can become limiting for this statistic (depending on the number of events collected and the number of parameters compared). Therefore, a reasonable number of bins is roughly 10% of the event count – leading to a minimum of about 10 events per bin.  The number of bins can be changed at the bottom of the comparison window.

As the number of parameters being compared increases, more events may need to be collected in order to distinguish subtle variations in the populations. More events will lead to better precision.  (Inclusion of parameters in the comparison platform which Do Not vary between populations does not degrade the ability to distinguish the populations.)

Note that the computations in the Population Comparison platform are memory intensive. You may need to allocate more memory to FlowJo (more information on memory requirements).