The Daily Dongle

7.5 Candidate Release 2 (May 5th update)

Please Note: this beta is intended to gather early feedback about some changes we've made in an effort to improve performance and reliability of the Windows version.

Supported Operating Systems:

• Windows XP and Vista, 32 or 64 bit.
• OSX Leopard or higher (64 bit Core2Duo only, no java 1.6 exists for coreDuo or older.)
• Linux (tested on Redhat and Ubuntu..)
  

Get the program for your matching OS here:

Windows:  download the installer and double-click it, a install wizard will guide you thru the process.

OSX: Download Flowjo then extract it to a folder, and run fj.sh from terminal (syntax: "./fj.sh" )

Linux: launch it via the webstart link

Full version/fixes history: 

FlowJo on Linux - update

** update 2/20/09 **

Ryan posted a webstart launcher for FlowJo 7.5 for Linux computers. We tested with Ubuntu 7, 8, and the current redhat version.

Installation:
- obtain and install the Java Runtime Environment for Linux 1.6, along with Java Web Start
- Start the program with http://umpqua.treestar.com/webstart-linux/flowjo75.jnlp .

Current support of FJ on Linux is handled by Ryan, whom you can reach via the address linux@treestar.com

Thanks to Ryan, FlowJo now runs on Linux.

Loadin', scalin' and compensatin' accuri FCS3 data files

another movie post. Subject: how to comp-wiz in Flowjo 7.5 b99 and how to 'handle' accuri data.

EOTD - DiVa/FlowJo Scaling.

This is to address a question from Purdue Cytometry Mailing List:

I Exported my 5 decade flow data from DiVa into FCS3.0 files. In FlowJo, it appears that the data is showing in 4 decades, rather than 5, with the last decade being squished against the axis. What's going on?

Historically, when we get this question, there are two causes, each discussed below, with a solution:

1. the data was actually exported as FCS2 files. To check if this is a case, open one of your exported FCS files in MS Word, Notepad, or any other text editor. You will see pages of junk, but the first few paragraphs should be in plain text. In fact, we only need to look as far as the first 4 characters. For example, here is the first few lines of a file I made in DIVA 6 :

FCS2.0         256    1853    1854   24425  0  $FILCompensation Controls_CD3 Alexa 700 Stained Control.fcs$SYSWindows XP 5.1$TOT1254 CREATORBD FACSDiva Software Version 6.0

as you can see, it's really an FCS2 file. There is a setting in DiVa you can toggle to choose between FCS2 or FCS3 flavor in your export.

2. the data is truly FCS3, but FlowJo's 32 bit preferences are set in a way that forces the data to be shown in 4 decades. To change this on a...

Mac V8: Click FlowJo menu, Preferences, Worksapce tab. Bottom right, notice "Define..." button. Click it to open 32 bit file prefs. Notice the "Pulse Height/Area" text fields - make sure these are set to 5. Close prefs, and remember to quit/reopen FlowJo for the changed pref to kick in.

PC V7.2: Edit menu in Workspace Window has the "Edit 32 bit prefs.." option - click it to get to the panel, set Pulse area/height to 5.

PC V7.5: there is a "FCS3" tab in prefs now, where you can set pulse area/height to 5 decades.

new betas posted

7.5 b86 - with some fixes in the workspace (gate dragging behavior, group dialog still somewhat broken ) and layout editor. Get it on the beta post here on the DD (scroll back a few posts.)

8.8.5 b1 - just posted, fixes a bug with exporting and an obscure crash with "summary files". Get it here.

Workaround for 8.8 File Export Bug

For those of you who would like to export your files out of FlowJo (for Mac) as new .fcs files, you may have come across a nagging little bug: your file names ($FIL) get cut short and no longer make sense. A workaround is presented here to ease the headache.

The key? Assign a separate set of keywords that you can use after exporting to orient yourself to the proper sequencing of your sample set. To save time add these keywords at acquisition, but in case this step was passed up, the example below will show how to add the the keywords in FlowJo.

To add the keywords in FlowJo (for this example, Tube Name and Tube Number), go to Workspace=>Add Keyword. A new Workspace column will appear with each keyword added. Type the keyword into the first cell and then copy/paste that word into each cell in the column (alternatively, use a spreadsheet to copy/paste an entire column of cells into FlowJo).

If you are going to use keywords you assigned at acquistion, throw them into the Workspace from by adding columns from the Edit button.

Then create a value series (in this case, 1 to 5) by typing in 1 (or any other number) in the top cell of the Tube Number column. With your cursor still in that column, go to the Workspace menu, hold down the Option key and select Create Value Series from the options.  You can then create a numerical value series however you would like.  1-16, 100-150, etc.

After doing these steps, my workspace has the two columns I added: 

 

Change the name that FlowJo displays in the workspace "Name" column in preferences (go to Workspace=>Workspace Preferences). If you already set up keywords during acquisition (best practice), it is very easy to select a combination of keywords that you would rather have displayed instead of the $FIL keyword (aka data file name). In my workspace, I made two new keywords with named Tube Name and Tube Number.  I can then export everything with the file name Tube_1 through Tube_5.

After exporting as a new .fcs file (Workspace=>Export) I added the samples back into FlowJo and here is what is displayed in the Name column:

 

Granted this isn't the most elegant solution, but the "frustration savings" should more than offset the cost.

For more on Exporting, go here: http://www.flowjo.com/v8/html/export.html

---

content credit: Nicholas Ostrout
editing credit: Ryan Farmer

Guava FCS files

> Is there any information opening data collected on a Guava
> cytometer in Flowjo?  (the compound FCS files)

FlowJo's founders have decided not to support the Guava format some time ago. I advocate the need for this periodically but this one has been outside of my zone-of-influence ;). You can help by griping  directly to flowjo at treestar dot com - which gets read by my seniors. FlowJo development is user-demand-driven. Be sure to point out that FlowJo V8 itself creates and supports concatenated multitube files, previously described elsewhere on my blog.

If you or someone in your group has programming savviness, you could make your own utility to extract the files from the concatenated one to flowjo-friendly one-tube-serving-size files. The header of the original concat'd file probably has the binary offset indeces for all the tubes included (TOC  if you will). Find the offsets, go to them, grab the data there, make a new header based on that data (mostly copy the original one except for the offsets/checksum values) and write
the header + chunk to a new FCS file. I don't have a "rosetta stone" for the specific nuances of Guava FCS format, but if you contact their engineers they might help.

FJ question from Purdue list

This is to address a recently posted FlowJo question on PCML:

does anyone know if it is possible to scale one histogram to match another in terms of number of events acquired?

There are three ways to go about this: export portion of events to a new file, gate on time or event # parameter, or normalize the overlay without fudging the data.

Exporting - if you care about uniform sampling - Take the 10K event file and export it un-gated to a new FCS file, with an option to only include 7000 events (which flowjo will randomly sample from the 10K).

• Mac users: click on the file in workspace, then click Workspace menu, choose export, and specify the # of events.
• PC users: I recommend 7.5 beta rather than 7.2.x for this task, link in a different DD post

If you don't care about the sampling bit, you can try to create a gate on the Time parameter (if you don't have time parameter saved, use Event number, which is enabled by a preference in FlowJo). Problem with this approach is that our binning of the data creates big "jumps" in between movements of a gate, so an exact count of 7000 may not be possible.

• Mac users: enable Event # parameter in Preferneces, look in the Graphs tab for an area called "Axes Options" then check the pref that reads "Add Event .."
• PC users: Graph window's Display menu, "Show Event..", then Edit/Save state to prefs.

The easiest method, in my opinion, is to change the Y axis scaling of your overlay to normalize for peak heights. Since Y axis on histograms is irrelevant, and distributions are relative, there is little statistical, and no visual reason, to match event numbers exactly.

• Mac users: double click overlay, check the "auto" option for Y axis.
• PC users: double click Overlay, switch to % max (normalized by peak height) or Unit Distribution (normalized by Area Under Curve)

I know it is possible to change the y axis to display # cells events on a simple overlay at the layout, but can't seem to find a way to do it for the histograms generated from the population comparison platform.

The population comparison was developed to have the capacity of comparing different size populations, and to make that easier, we normalize the populations to peak height and plot them on a "%" scale. Currently the Mac version will appear to label the Pop Comp plots placed in the layout with an absolute (#Cells) scale, which seems like a bug we should fix. I don't know if there is a statistical or otherwise compelling reason to do this, so if you want to let us know your thoughts, email flowjo at treestar dot com, or leave a comment below (anonymous OK :) but please be constructive. )  

Good thing to do when upgrading FJ from older version (OSX)

1. Search for "flowjo" where type=application, and remove all found versions (I found 18), then empty trash.
2. Download the file http://umpqua.treestar.com/bin/FlowJo.8.dmg (static link, always up to date)
3. Open the volume and drag app to Applications
4. Right-click on old workspace, choose Info:

5. Look for the "open with" section, and point it to the current version. If it's not listed, click Other, and navigate to Applications, then select latest version of FlowJo:

6. Click "Change All" - this will ensure all your .jo files will open with latest FlowJo version.

7. Rinse & Repeat for .jot files.

does number of vertices in my polygon gates matter?

Admittedly, I don't get asked that question a whole lot, but once I did, it sparked an interest, and I put it on my todo. Today I got a round tuit, and have some benchmarks to show yee. First I whipped out a quick titration spanning gate complexities of {3,20,40,80} on a few data files ranging in size form 10K to 2.5M events:

Of course, most of my gates fall in between 2-20 vertices, so how does that scale on a 11M event (500MB) file?

(X axis - number of vertices, Y axis - time to calc gate in ms)