$PnG keyword in FCS3 spec
rhicks at cytekdev.com
Wed Nov 19 05:42:47 EST 2008
Not a founder or guru (not for want of trying to get on the committee), but
I'd tend to agree that the gain should be used to normalize linear
parameters, particularly in the case where data are acquired in linear on a
low resolution scale to improve quantification (by side-stepping log amp
errors)and where the range of intensity between samples requires that
different gains be used to ensure all samples are on scale and properly
measured. This removes the onus on the experimenter to manually note the
gains used on each tube and then adjust average intensities post-analysis (I
found it annoying to have to do this on the rare occasions that we used
linear acquisition in this way).
With the higher dynamic range/resolution on modern machines, this may not be
so necessary since acquisition on an 18 bit scale allows the equivalent of a
10 bit scale with PnG up to 256, but people may still prefer to move a peak
around so that it falls into a similar number of bins, then normalize using
the gain (rather than compare a squished low peak straddling channels 100 to
200 with a brighter peak straddling channels 100000 to 200000) to obtain
data that doesn't have artefactually low CVs due to binning at the low end
of the scale (Jim Watson touches on this in his data analysis book I think).
One drawback of this approach is that it can lead to surprising full scale
values (eg 1.023 on a 10 bit machine using a linear gain of 1000 on FSC) -
meaning that you have to adapt your tick mark algorithms to draw fractional
tickmarks, or just have the first tick mark at the extreme right.
I'm not sure that any manufacturer exploits the $PnG info in their plotting
so for playback consistency it may be worth having a default "ignore $PnG"
preference that the user could override to display data scaled according to
It's also possible that in some cases for log data that the linear gain is
not applied, though a value is recorded in the header whereas in other cases
it may be applied (eg FSC coarse versus fine gain on Calibur), and $PnG may
not be an simple multiplier at all as in FSC "voltage" in DiVa software -
leading to some confusion for the programmer and the user.
Tel: +44 (0)208 1337 968 Fax: +44(0)208 5889 004 Skype: ray.hicks.cytek
Cytek Development Europe, Unit 8 The Maltings, Millfield, Cottenham,
Cambridge CB24 8RE. UK
From: maciej simm [mailto:simm at treestar.com]
Sent: 13 November 2008 19:26
Subject: $PnG keyword in FCS3 spec
there is some internal debate here over the usage of the $PnG keyword.
As per the FCS3 spec, it is not a required keyword, and is defined as..
*$PnG/f/ $P2G/10.0/ - *This keyword specifies the gain that was used
to amplify the signal for parameter n. This example shows that
parameter 2 was amplified 10.0-fold before digitization.
so my question to you FCS3 spec founders/gurus: If my hypothetical
parameter P1 has a $P1G=20, $P1R=1024,what are the min and max values of
the parameter assuming linear scale:
a) 0-1023 since $PnR is more important
b) 0-50 since it was multiplied, hence we have to divide R by G
thanks for any comments!
FlowJo Support & Engineering
More information about the Cytometry