RE: rare events

From: Qi, Hai (haqi@utmb.edu)
Date: Fri Dec 03 1999 - 23:12:23 EST


Ann Atzberger and others,

Maybe I did not explain the original question well. The real case was: I had
10 million lymph node cells and I mixed it really well by pipetting and
vortexing. Then, TWO samples, each with 1/5 million cells, were aliquoted
into TWO tubes and run separately. One gave me 0.04% and the other 0.15%. If
I assume the best mixing was achieved (a plot of fluorescence over time
shows whether positive events are well distributed) and dead/debris was
minimal (my cells were 99% live), I am still bound to be worried by
something else, the systematic error of sampling. This is just like
epidemiological survey: I can only survey a certain number of people (cells)
to represent the total, all people (all 10 million cells). It is a
statistical problem to determine how big that number would be so as to gain
certain (but never 100% unless the total were surveyed) confidence on the
representativeness of my results. Dr. James Leary's paper (Cytometry 27:
233-238, 1997) addressed this issue and there is a link to a lecture note by
Terry Hoy, http://www.uwcm.ac.uk/uwcm/hg/hoy/rare.html
<http://www.uwcm.ac.uk/uwcm/hg/hoy/rare.html> , provided by Jeff Barry.

Thank you all for replying to my message. 

Hai Qi
Department of Pathology, UTMB
Keiller Bldg., Rm. 1.104
301 Univ. Blvd.
Galveston, TX 77555-0609
Tel:  	409-7728163
Fax: 	409-7476869



This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:54:18 EST