As Jose and Julio point out the data is
readily available in the form you are looking for.
A wider question is what you take measurements
to mean.
These coefficients need to be qualified by
the area each fluorophore occupies in the ROI: if each occurred in 50% of the
pixels, then chance alone would produce an overlap of 25% (uninteresting), while
overlaps of say 10% or 40% would be more interesting – a non random
mechanism.
If say they are both water soluble they
might well appear throughout the cytoplasm and show 100% colocalization (using
some variant of overlap).
Also bear in mind that these coefficients
are very sensitive to the threshold used to decide if a fluorophore is
present/absent in individual pixels.
A more interesting measure is correlation –
does the variation in the intensity of one fluorophore match that of the
second. This shows whether there is some underlying linkage between the distribution
of the fluorophores. The variation would be caused by inhomegeneity in the
volume of distribution. A correlation measurement for our two fluorophores that
simply appear in the cytoplasm will be very low.
In practice use both overlap and
correlation when considering colocalization and, as with all measurements,
always consider exactly what has been measured and how it might be misleading.
Jeremy
Dr
F451a
Cell Biologi
Wenner-Gren
Inst.
The Arhenius
Lab
S-106 91
From:
Sent: den 5 februari 2009 02:13
To:
[hidden email]
Subject: Re: colocalization
question
-
Actually, many colocalization programs, including the imageJ plugin
"Colocalization finder" will tell you the percent of each channel
that is colocalized. see for instance:
http://rsb.info.nih.gov/ij/plugins/colocalization-finder.html
--
Julio Vazquez
http://www.fhcrc.org/
On Feb 4, 2009, at 5:05 PM, Jean-Pierre CLAMME wrote:
Hi,
When
you do colocalization measurement (in image pro for example) between let say
the red and a green channel the co-localization coefficient tells you how much
red and green are co-localized. But as much as I understood, it doesn't give
you information on the fact that for example red is always co-localized with
green but green is not always colocalized with red ? What kind of analysis
would I need to compare this between samples ?
Thank
you in advance,
JP
-
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Jean-Pierre CLAMME, PhD
Senior Scientist
Nitto Denko Technical
501 Via Del Monte
E-mail: [hidden email]
Phone: +760.435.7065
| Free forum by Nabble | Edit this page |