Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

Judy,

I'm going to stick my neck out an make a suggestion...

Co-localization is a correlation.  I suspected that statisticians can do a three 
variable correlation with minimal difficulty.  I was interested, so I googled 
around this idea with the following (half decent) results;

http://faculty.vassar.edu/lowry/ch3a.html

http://books.google.ca/books?
id=bmwhcJqq01cC&pg=PA1012&lpg=PA1012&dq=correlation+three+variables&so
urce=web&ots=I9IYZSS-kj&sig=mItL7gIHEWt3RuQydR7XY4-
PHsE&hl=en&sa=X&oi=book_result&resnum=4&ct=result

I think the short answer is, talk to a statistician.  As you mention, I have 
never seen this implemented.

One caveat that I suspect you are aware of... In my experience people tend 
to relate visual colocalization with physical association.  Very rarely do I see 
anyone consider a negative control, and then a comparisson of correlation 
coefficients between experimental and negative control to determine 
significance.  I suspect that if you went this far, and then mixed in a third 
variable (protein), it would be hard to obtain significance simply due to 
resolution limitations.  Of course, this would depend on abundance and 
distribution of protein(s) of interest.

Would be interested to hear thoughts if anyone has further suggestions.

Dan



------------------------
Dan Stevens, PhD
Cellular Imaging Specialist
Carl Zeiss Canada