FlowJo and Accuri's CFlow

I met today with Accuri's Jack Ball (CCO) and Clare Rogers (ApSci). We got a demo of the C6's software CFlow - made in java! - and spent some time talking about interfacing FlowJo with Accuri data. As some of the blog's readers are users of Accuri C6 units, I thought it would be good to summarize some of the points here.

 

The GUI, explained in FlowJo terms - similar to cellquest or DiVa, there is only one "group", in which all samples reside, there is only one "gating tree" and one "layout". Once you make a gate, it "applies" to all samples present in your experiment.There is only one place to arrange a series of plots.

We're going to work on reading the Accuri "experiment" files directly into flowjo (the data, not so much the gates/layouts etc.) Since the "experiment" file is just a zip file with a .c6 extension, FlowJo V7 is no stranger to archiving/dearchiving ZIP's -  the groundwork is already in place.

The new version of Cflow due out early April will have slightly different keywords, and we're working to make sure the next flowjo version matches these for:

- parameter scaling (gain, offset, min/max useful channel)
- parameter and meta labels.

Some of the remaining todo revolves about handling the extra decades found in accuri files. 16.7M channels, or 24 bits, or 10 decades, or James-Wood-Knows-How-Many Decibels, is more data than any monitor can natively display. 

Thus, when displayed, the data has to be slightly reorganized into "pixel-bins". CFlow gives you interactive zooming function. See how it works here on a scatter plot: each plot in the sequence below is being zoomed in on, from left to right. This redraw takes 2 mouse clicks and produces an instant update:

 

In FlowJo for Windows, there is no matching functionality yet (until 7.3) but the Mac version has a notion of setting min/max limits on a "derived parameter". So, using derived parameters platform, which takes 12-14 clicks per adjustment, you can "dial in" your range as shown below in the FJ plots. (the axis values on FJ have to be divided by 16.7 - something we're looking to resolve soon.)

 

The FlowJo result has picket fencing. I think this is because we bin the data once at load, and then simply "spread it around", whereas Cflow re-bins it with each zoom. Hopefully that's how we'll do it in the next version of flowjo...

An interesting question came up: should the user choose to compensate the data in FlowJo, would they derive the parameters before or after compensation?

If you derive first, and comp second, you end up with compensating re-binned data.
If you comp first, and derive second, you end up without a compensated parameter, so bi-exp won't work.

it's a tough balance - I haven't done this with accuri data a whole lot yet, but I think Clare and I will work on it.

Comments

[JM]

fyi

[Kuldeep]

This is an interesting read. Our lab has recently purchased an Accuri C6. Being used to analysis on FlowJo, I would like to continue my analysis on FlowJo. Would there be a tutorial on how to setup flowjo to analyze data acquired by the Accuri C6?
Thanks,
Kuldeep

[maciej simm]

Kuldeep, Clare Rogers (apSci over at Accuri) has created a PDF outlining the workflow of reading CFlow data in FlowJo for Mac.

The PC version is still pending fixes for certain things to work right.

[Rina Gendelman]

I've been waiting for the PC version that can handle data from Accuri C6 for some time. Do you know when it will be coming on line? Or, if you can recommend a compatible cell cycle analysis software that I can use until FlowJo for PC is ready to handle C6 files.

Sincerely,
Rina