RE: Sort purity

From: Gerhard Nebe-von-Caron (Gerhard.Nebe-von-Caron@Unilever.com)
Date: Thu Nov 08 2001 - 13:25:14 EST


Just to add to the comments of David and Jens . Apart from the fact that a few
things can happen to the fluorescet molecules of a cell between sorting and
reanalysing, the key thing to consider is indeed the problems of distributions.
The simplest test indeed is to sort the top 1/2 of a bead population of high
and very low intensity.
In cases were the variation of the signal is based on the distribution of the
fluorochrome, indeed the sorting will reflect that. However, at low signal
intensities one starts playing with statistics.  This gets even worth when
sorting compensated populations as the compensation increases the variation.
Jens refers here to the formation of doublets in primary samples. Another
related problem comes from coincidence events. Platelets can serve as a nice
example. When coinciding with leukocytes (or even being attached to them) they
tend to give fluorescent signals but do not show up in changes in light scatter
or signal pulse. Again after a high dilution this coincidence will vanish.

Regards
Gerhard
-----Original Message-----
From:	Jens Fleischer [SMTP:jfleischer@tesionmail.de]
Sent:	Monday, November 05, 2001 6:01 PM
To:	Cytometry Mailing List
Subject:	Re: Sort purity



In my humble opinion we have to consider various effects:

1. As David McFarland correctly described, our populations are most
often (always?, don't nail me on this, this would be another endless
thread...) gaussian distributions, thus if you sort a "histogram
shoulder" you will always get again a gaussian distribution and nothing
between "goal posts". Recently I saw a software package that nicely
calculated the possible gaussain distributions under an
immunfluorescence curve, similar as the modelling for cell cycle
analysis. I do not know when and how this will be commercially
available.

2. Als Volker Eckstein wrote, loss of fluoresence and bleaching can have
a serious impact, for example PerCP is quite sensible for this. He also
mentioned the gaussian distribution by telling that a cell never falls
into the same fluorescence channel if you measure it several times.

3. We should also not forget the formation of doublets in the primary
sample which fall apart after sorting and washing the sorted fractions.
Ideally, possible aggregates should be excluded from the sorting
process. However, this is often not possible using the traditional
scatter height signals. Depending on the orientation while going through
the beam intercept you cannot resolve all aggregates by comparing the
height. Optimally, one should use the area signals also for scatter AND
fluorescence.

4. Electronics: I do not think that there are really bugs in the sorting
electronics of any vendor (we would definitely have realized it before).
If your machine can sort beads 100% pure there is no reason why it
should not be able to do this with cells. So sample preparation and
sorting conditions (sample/sorted fractions warm/cool, flow rate above
threshold, sort rate, abort rate, sorted drops, coincidence) play the
most important role.

And by the way, don't forget to do some extensive washing before you
reanalyse your sorted cells again on the very same machine you sorted
them. The sample tubing can be a steady fountain of cells after three
hours of sorting...


Jens Fleischer



This archive was generated by hypermail 2b29 : Sun Jan 05 2003 - 19:01:38 EST