ICA for Triple Labelling

[alexa_11]

Hi,

I am attempting to find a method of retrieving an ICQ for three channels using the Intensity Correlation Analysis plugin. One of these channels is a marker for synapses, and I'm interested in looking at the colocalization of the other two proteins with this marker.

This might be a bit far fetched, but I was wondering if it would make sense to use the +PDM image (with all the negative pixel images removed since they are blank in my case, and crop the extra black strip added to the bottom so the size corresponds with channel 3) and run it through the ICA program with a third channel. I know it works, but I'm not sure if it makes sense to take the ICQs as given.

Thank you,

Alex

[Daniel James White]

Hi Alex,

On Apr 6, 2011, at 6:00 AM, IMAGEJ automatic digest system wrote:

>
> Date:    Tue, 5 Apr 2011 09:19:57 -0700
> From:    alexa_11 <[hidden email]>
> Subject: ICA for Triple Labelling
>
> Hi,
>
> I am attempting to find a method of retrieving an ICQ for three channels
> using the Intensity Correlation Analysis plugin.

then you need to do each channel pair individually,
so thats 3 tests.  

> One of these channels is a
> marker for synapses, and I'm interested in looking at the colocalization of
> the other two proteins with this marker.
>
> This might be a bit far fetched, but I was wondering if it would make sense
> to use the +PDM image (with all the negative pixel images removed since they
> are blank in my case, and crop the extra black strip added to the bottom so
> the size corresponds with channel 3) and run it through the ICA program with
> a third channel. I know it works, but I'm not sure if it makes sense to take
> the ICQs as given.

Can you imagine or write out how this might make sense mathematically?
If you can, then please share.
If you can not, then you are not doing something that might give any meaningful results?

Think about how you are framing your question.
Coloc as defined by the ICQ method of Li
is able to compare 2 images, not three or more.

you can talk only about the correlation between pairs of image data sets.

I am not sure sure what toy are trying to achieve???
Can be explain in more details what you are trying to measure exactly?

cheers

Dan



>
> Thank you,
>
> Alex

Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Visualisation, Processing and Analysis
Light Microscopy and Image Processing Facilities
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany

[phone] (German Mobile)
[phone] (Work phone at MPI-CBG)
[phone] (Fax MPI-CBG LMF)

http://www.bioimagexd.net  BioImageXD
http://pacific.mpi-cbg.de                Fiji -  is just ImageJ (Batteries Included)
http://www.chalkie.org.uk                Dan's Homepages
https://ifn.mpi-cbg.de  Dresden Imaging Facility Network
dan (at) chalkie.org.uk
( white (at) mpi-cbg.de )

[alexa_11]

Hi Dan,

I am looking at two proteins that localize to synapses however they can also be located elsewhere, for this reason I need to compare with the synaptic marker.

I already analysed the images as pairs x3 but it would be great to see in which cases protein A colocalizes with protein B in the synapses. With the analysis of each pair you are able to see how often protein A or B is localized to synapses, and how often protein B colocalizes with protein A, but you are not able to quantify the three as a whole.

The way I understand it, the +PDM image gives you the pixels which colocalize in a dependent manner (in this case giving me the population of protein A that localize to synapses).

Now, I'm not sure if the pixel intensities in the PDM image would be of any meaning when comparing with protein B. And mathematically, I'm also not sure if this makes sense.

Thanks,

Alex

[EliseStanley]

A three dimensional ICA is, I think, a possibility but I don't think I'm up to the mathematics either!  However, I don't think it would really help here unless first, your presynaptic terminals occupy more than a number of pixels in the image (some are at the limit of light resolution so less than one pixel) and if the third (presynaptic) marker stains all regions of the terminal.

there is, however, a nugget of cleverness here.  The problem with many studies that do an ICA analysis is that they include areas within which the protein can not exist - the most notorious is both intra- and extracellular regions.  What happens then is that a positive ICQ is almost certain as even if the proteins are random inside the cell there is some probability of colocalizatoin whereas there is no probability outside - thus, when summed the ICQ value will almost certainly be positive.

What would be neat (and I think this is what you were after) is that if you could stain all regions of your image that were 'common' (such as intracellular) you should be able to do an ICA analysis on those alone and that would be valid.  I think this should be relatively easy as you can threshold the third stain as positive (+1) or negative (0) and multiply the PDM value - since (if I remember right) 0 values are removed during analysis.

Elise
PS please see topic today on the ICA 10 year anniversary!