Summary Document

This document is a summary of message from the Purdue University Cytometry Laboratories Cytometry Discussion list.

http://www.cyto.purdue.edu/hmarchiv/index.htm

The original post:

Jamie Barber
AHRC Cell Imaging Core Facility
Dept. of Infectious Diseases
Univ. of Georgia, College of Veterinary Medicine
111 Carlton St. Bldg 1077
Athens GA 30602
Office: (706) 542-4092
Laboratory: (706) 542-9862
barber@uga.edu

We are currently doing some proliferation assays using CSFE to follow divisions. We’re acquiring the data on a digital platform, an LSR-II, running FACSDiVa. Diva’s biexponential plot scale gives us beautiful peaks in the lower CSFE areas. However, since the lower logs are compressed, I’m not sure if the pretty pictures are valid for analysis (are the lower populations lumped together?). Any thoughts?

Secondly, if this is a reasonable way to look at proliferation data, any ideas as to how to “export” a biexponetially-scaled data set to Modfit??? Am I asking the impossible (I think I am)?

The responses:

Maciej Simm from Treestar (FlowJo Technical Support) wrote, “The question comes by my desk a few times a month, and I usually just tell them to turn down ‘width basis’ (something you can do in FlowJo, I don't know if other softwares do it)”
Monisha Sundarrajan from BD Biosciences wrote, “Yes, to analyze the data in Modfit, exporting bioexponentially scaled would not work as bioexponentially scaling is a characteristic of Diva and once you import the data it is no longer is presented in a bixponential fashion.
Yes, the bioexponential scaled data is compressed at the lower log but it is real data; it is just presented in that format to give a better appearance, to give the population a spread out appearance, however you could compare the statistics to give you a fair idea if the population is real or not.”
William Schott of The Jackson Laboratory wrote, “Only the display is transformed, not the FCS3 data itself, so another program should be able to read the files with no problem...”
For Modfit LT 3.x: A beautiful solution for using FCS3.0 digital data was presented by Mark Munson of Verity Software House (makers of Modfit). Thanks Mark, it worked great! This method made for inclusion of the lower-value data that was otherwise cutoff and discarded during Diva to FCS2.0 conversion:

1. 1^st, be sure to download and install SP3 for Modfit LT 3.x from the VSH website.

2. Export data files as FCS3.0.

3. Convert digital data to log-scale.

4. Analyze for proliferation.

5. Follow Mark’s instructions, below (screen captures removed):

o “[To] convert a parameter to log in LT 3.1 SP3. When the file is first opened, the same dialog that asks you to choose the parameter [for analysis] also allows you to convert the parameter to log...

o When clicked, the Convert button opens the log conversion dialog. Check the parameter to convert, and then choose how many log decades to use. The default is 4-decades, which works well with the default proliferation model.

o However, you may alternatively choose to convert the data to "computed" number of log decades, which for 262,144 resolution is about 5.42 decades.

o If you choose "computed", then be sure to tell the Prolif Wizard to use 5.42 decades, which makes the spacing 15.35 channels.

o The analysis will now run at the full range of the data using the adjusted spacing.”

Thanks to you all!