Re: Purity??

From: Derek Davies (daviesd2@icrf.icnet.uk)
Date: Thu Feb 03 2000 - 12:35:36 EST


Hi Carl-Magnus,

If I am right in interpreting what you are saying, you are sorting the
lower part of a positive peak and the re-running it to check purity, is
that right? The general way to assess purity would be to re-run the
sample on the machine you sorted on, leaving all parameters unchanged.
When you sort a defined population, say, double positive, you would want
most (>90%) of the cells to fall within the original sort gate, but I
would accept that there may be some drift (generally toward the 
negative); as long as there isnt significant contamination with other
populations though I would be happy.

However, I have often done sorts on, for example, integrin-stained
keratinocytes where the user wnats to sort the top and bottom 10% to see
if they have different growth characteristics. In this case what they
want to see is a precipitous cut off at one end of the re-run sample.
However this is unlikely ever to happen and you will see what you are
seeing. This stems from the fact that if a cell goes through the machine
twice it will not necessarily be in the smae place but its position 
will have a gaussian distribution so the cells at the edge of the sort
window may well go over it on re-running. In these cases either 'sort
purity' is meaningless or 'bottom x%'is meaningless. What I tend to say
is that we have sorted the bright from the dim but let the user know the
caveats. 

Hope this helps (or, indeed, makes sense!)

Derek


On Wed, 2 Feb 2000, Carl-Magnus Hogerkorp wrote:
> How does one define sort purity? Is there any consensus around how to
> present purity figures?
> 
> To exemplify:
> If sorting a CD34+/CD38- "population" of 6% (based on the 34+ of the total
> viable cell population) and knowing how the machine works, a population
> rather corresponding to the ~25% lowest gets enriched. Within the original
> sorting gate approximately 60% of the events fall and the rest will fall
> above the sorting gate. These 40% are not something I would define as
> impurities since they are a part of the sorted cluster of CD34+/CD38- cells.
> How do I define sort purity? Is it 100% of the 25% lowest or 60% of the 6%
> lowest.
> 
> It becomes pretty obvious that, if wanting to sort these very 6%, the sort
> gate should include perhaps the 1% lowest or something in that way. However,
> most users of sorted fractions are not aware of the limitations of the
> sorter but merely ask for the 6% lowest. This of course is a bit of a
> dilemma for the person operating the sorter since there are no simple
> algorithm to apply correcting for this phenomenon. How do others deal with
> this?

************************************************************************
Derek Davies                       Voice: (44) 0207 269 3394
FACS Laboratory,                   FAX: (44) 0207 269 3100
Imperial Cancer Research Fund,     e_mail: derek.davies@icrf.icnet.uk
London, UK

Web Page: http://www.icnet.uk/axp/facs/davies/index.html

In tenebris lux 							 
*************************************************************************



This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:55:29 EST