Re: Data Management

From: Todd A. Johnson (tajohnson@UCSD.Edu)
Date: Mon Jun 10 2002 - 13:43:44 EST


Jeannine-

I have been working on research databases for flow cytometry for the last
couple of years and would suggest using CellQuest's export feature rather
than hand entering your values.  I personally prefer FlowJo's analysis and
export features, but under CellQuest's File menu you will find a function
"Export" that will export all of the statistics in your file into a tabbed
text file that can be opened in Excel.  I believe there is also a checkbox
for having data appended to an existing file.  If I remember correctly,
there is also a batch processing function, such that you could have a
template for analysis run through your samples and perform this export
automatically as each new file is analyzed.

The problem that I saw with CellQuest was a difficulty with labelling a
particular row of data with something meaningful.  As far as I can tell
from looking back at an export we did a couple of months ago, each separate
row of statistics data that you see in a CellQuest analysis, i.e.
UL,UR,LL,or LR from quadrant analyses or M1, M2 statistics from histograms,
will be exported as one row into the spreadsheet file with all of the
associated header information attached to the beginning of the row.    The
old CellQuest export file that I'm looking at has in one row all of the
following information:

File Name/Sample ID/Patient ID/Patient Name/Case Number/Acquisition
Date/Gate/Gated Events/Total Events/Label/% Gated/% Total/X Mean/X Geo
Mean/Y Mean/Median

such that you have in just one row all of the last six parameters as data
values pertaining to the one "Label" (UL,UR,LL,LR,M1,M2...). but also no
columns that describe the stains that were used.  In our analysis file, we
had seven sets of stains and while working with CellQuest,  the only thing
that  I tried was exporting the whole report that the clinical flow lab had
provided to us.  In retrospect, I would probably make a new analysis
template for each set of stains (i.e CD3/CD19/CD8/CD4 or
CD45RA/CD62L/CD8/CD4), and export the data into separate stain specific
files.  In the assumed absence of having CellQuest export a meaningful
label for something like a quadrant statistic or a histogram marker, this
approach could facilitate relabelling the data with something useful(i.e. a
row of data from a dotplot of CD3/CD19 should be labelled as either
"CD3+CD19-" or "T cells" rather than just the "LR"  that CellQuest
uses).  You might also narrow the analysis even further and make an
analysis template that just exports one dotplot's quadrant analysis or a
histogram's statistics so that you can go back in Excel and do a find and
replace in the label column; "T cells" for "LR", "B cells" for "UL"...  You
might need a lot of separate analysis templates, but then could recombine
the data into one file after a little post-analysis processing.  There's
always more than one way to get the job done...try the export on a file,
turn off some of the parameters that are exported, redo the export, and
compare the resulting files.  Playing around like this might give you a
better idea of what labels, i.e. File, Patient Name,Gate, you really need
for each row that is exported.

Feel free to e-mail if you need more detailed information.

Hope this helps!

Todd



At 05:18 PM 6/6/2002 -0400, you wrote:

>Hello to all the helpful, and far-more-knowledgable-than-me flow
>cytometrists out there:
>
>I've been involved in what started out as a relatively small
>cytometry-based human study that has recently ballooned into something
>much larger (10-fold increase in study participants, and still
>growing).  I have been taking my flow data, hand-entering it into
>spreadsheet databases, which get sent to a statistician to correlate
>with other clincal features of the diseases we are studying.  What
>started out as a relatively simple task (typing flow values into an
>Excell database) has become more and more time-consuming.  My question
>is for those of you especially who run clinical flow cytometry labs that
>generate high volumes of data:  How do you manage the data once it is
>generated?  Are there any programs available that take the data
>automatically once the samples have been run and enter the values
>directly into a spreadsheet?  Or do you hire work study students to do
>this for you??  Please help me!
>
>I'm running a FACSCalibur, and CellQuest software.
>
>Thanks a bunch!
>
>Jeannine Navratil
>University of Pittsburgh

Todd A. Johnson

Computer Systems Manager
UCSD, Department of Psychiatry
Dimsdale Lab
CTFA 309/ MC 0804
(619) 543-2504
tajohnson@ucsd.edu



This archive was generated by hypermail 2b29 : Sun Jan 05 2003 - 19:26:12 EST