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

Adler et al. (Replicate-based noise corrected correlation for accurate
measurements of colocalization. Jeremy Adler, Stamatis N Pagakis, and Ingela
Parmryd. Journal of Microscopy 230(1):121-133, 2008) suggests using
replicate images to estimate errors due to Poisson and background noise to
derive a correction factor to the Pearson correlation coefficient.

They also note the distinction between selection of a ROI in a scattergram
(which can produce some very high correlations), for example Zinchuck,
Zinchuck and Okada, Acta Histochem Cytochem _40_ 101 (2007), and selection
of a ROI on the image (which may be more meaningful).

I too, have concerns about the arbitrary thresholding and intensity
saturation in the Zinchuck et al. publications as expressed by Dan White
earlier in the thread.

I am hoping that the error correction described by Adler et al. can be
applied to the Manders coefficients in a similar way. Is anyone
statistically confident enough to venture an opinion?

If so, replicate-based noise corrected correlation, combined with the Costes
method ( Costes, S.V., Daelemans, D., Cho, E.H., Dobbin, Z., Pavlakis, G. &
Lockett,S. (2004) Automatic and quantitative measurement of protein-protein
colocalization in live cells. Biophys. J. 86, 3993­4003. ) of objective
thresholding and choosing image ROIs may well the current best practice in
quantitation of colocalization.

Kingsley Micklem

-- 
Medical Informatics Unit
Nuffield Department of Clinical Laboratory Sciences
University of Oxford
Room 4A12A, Level 4, Academic Block
John Radcliffe Hospital
Oxford OX3 9DU
(44) 1865 220555