BIG files

Benjamin C. Hunsberger 75450.167 at CompuServe.COM
Fri Jan 7 09:09:15 EST 1994

Don, Ben,
  You may have seen this dialog on the 'net.  I have wondered about this
myself.  When I collect listmode files with > 20,000 events, I never get all
the events back when I replay the FCS-converted file using either WinList of
Modfit.  The most events I've ever "recovered" is about 19,000.  Is there
some way to circumvent this?

Thanks, Marc

About your only seeing 19,000 events in WinList.   Your question needs more data
to answer.  First, you should click the Keywords button when loading the file to
see what the value is for $TOT.  It may be that the file only stores 19000 even
though you ran 20k because of a "stop" you put in.  The second thing to do is to
pull up a single parameter histogram, put a region across the entire width of
the x-scale, show stats for that region, and re-acquire the histogram.  The
number of events in the results window will be the total number in the listmode

If you are talking about "seeing" all the events, you need to click "show all

For very large files.  At this time WinList can not have a PACKET size greater
than 32K.  It CAN however display all events for a few 2 parameters at a time.
The display of all 1,000,000 events in a single 2p dot plot will consume
4,000,000 bytes of memory - each event stores an X and Y 2-byte value.   This
tranlates to: If you have enough available RAM memory (in this case I would want
16MB or more) WinList CAN handle these large files.

Relating to the size limitations on FCS files, the FCS header section describes
offsets to various parts of the data file.  Each offset can be an 8-byte ASCII
number, therefore the largest offset possible is 99999999 (99 million).  It is
very possible to exceed this limit with large acquisitions, esp. if lots of
parameters are stored.  For example, if 1 million events are collected with 6
2-byte parameters for each, you end up with a data storage size of 12 million
bytes. I belive this is being addressed by the FSC committe.


More information about the Cytometry mailing list