Warning on Coloc2

HI All,

I learnt about ellen arena from the imagej community as i wanted an answer for coloc2. I recently started using coloc2 and got a warning message saying that the y-intercept is far away from zero. Try using ROI and subtract background option. I am unable to understand how to circumvent this problem ? Which channel to choose for ROI and how to background subtract this fluorescent image.

Thanks in advance,
Rohitesh

1 Like

Hi Rohitesh,

here is a nice thread with an explanation from the developer of Coloc2:

Cheers,
Christopher

3 Likes

Hi christopher,

I was not able to implement correctly the methodology described in the article. Could please help me with steps for an image that i am attaching here.

Thanks,
Rohitesh

Hi Rohitesh,

the provided format (.jpg) is incompatible with image analysis.
.jpg uses compression read more here: https://imagej.net/Principles#Why_.28lossy.29_JPEGs_should_not_be_used_in_imaging

With what particular operation do you have problems?

Cheers,
Christopher

1 Like

Hi Christopher,

I am getting warning message on coloc2 stating that the y-intercept is too far from zero. I have tried subtracting background in diff ways but still the error appears. I have also included an ROI for channel 1 or 2 but the warning persists. I am sharing here another format for the same image. If any steps can be shared, that will be great.

Thanks,
Rohitesh

@rohitesh13

Let’s start by getting some details of your analysis first. What is your goal for this analysis? What do you wish to measure? For which channels?

This older thread post has some helpful links to past threads/posts with suggestions/help that is relevant to your case (read through them all):

In particular - read @chalkie666’s response here (also linked above by @schmiedc):

His response there is applicable to your case:

Hi,
I want to measure the colocalization between these two channels. I have tried using the approach that was mentioned in the links but couldn’t really resolve the problem. The warning message still appeared. May be I am doing some mistake here, I don’t know.
I will explain what I did. I split image using color -> split channel and then select green and red channel images. Then I created a square ROI on green channel and measured “mean” value using analyze -> measure. This I subtrated from the whole image using Math-> Subtract option. The same method was applied for the other channel. Post this, I started coloc2. But, the warning message still appeared. I don’t know if the values that I am getting are skewed or I can still go ahead with the colocalization values inspite of having this warning sign.
Please let me know. Thanks,
Rohitesh

Before you dive into the technical details you should answer this

It is understood that you want to measure colocalization, but maybe spell out the biological question before trying to solve the issues with Coloc2. The measures calculated by Coloc2 might not be what you are looking for. Also, as mentioned by @chalkie666 in the referenced thread, make sure you have positivie and negative controls (taken under the same imaging conditions). Without those, colocalization numbers are meaningless.

Once you are clear about this, you can follow the other advice: restrict your analysis to regions where the biology is happening (either by cropping or setting ROIs).

2 Likes

Thanks for your input. Prolly, I will spell out my biological problem here.
I have used two different antibodies in this image (no doubt about that). In a negative control or KO cells, same antibodies are used to test if there is any overall difference in the overlap between the two markers. Essentially, I want to show thru quantitation that there is a robust difference between the control and KO cells for these markers or antibodies. Since there is a negative control, so definitely one can determine the difference in overlap. Does it makes sense ?

Guys, any help on this matter will be much appreciated ? Awaiting response. Thanks!

Hello rohitesh13,
I’m sorry to butt in your conversation, but do you have a copy of your original image? There seems a relatively simple way to determining what you are researching other than Colocation.
I’ll see if I can assist,
Bob

Howdy,
Your co-located cells appear as yellow in the attached image.

Maybe this helps.
Bob

Hi Bob,

Thanks for this image. But, how did you get this output and what is the overlap index like ?

I am eager to learn.

Thanks,
Rohitesh

Rohitesh Gupta

Hello
I deeply apologize for the delay in responding. Scheduling cares for no man is an old saying.
First I must say that I am not a Biologist and therefore cannot guantee the data produced by my result is accurate. Please try to follow the logic used to determine if it will meet your needs.
I looked at this from a Photonics point of view so my first act was to measure the intensity of each color channel. I noted that the Blue channel, although more localized was much more intense than either the Red or Green and therefore interfered with the ratio between the Red and Green so I split the channels, subtracted the blue from both the Red and Green channels. I could not simply discard the Blue channel for as I stated it also interfered with the Red and Green. After the subtraction I then discarded the Blue because you seemed interested in the Red and Green colocation.
When comparing the Red with Green the Red overwhelmed the Green therefore I used Auto Adjust of Brightness/Contrast which would make the most intense of each channel equal and therefore when recombined into a single image the Green and Red with equal intensity indicated the colocation by becoming Yellow.
I did not analyze any Data other than the visual. I did not know specifically what you desired.
For numerical data you only have to split the Red/Green channels and use Process > Image Calculator > Difference as the operation. That will give you an image where the colocated cells are the DARKEST. So invert the image so that the colocated cells will be the BRIGHTEST, threshold and finally Analyze Particles to obtain count, location, etc, etc…
So, from a Photonics point of view it all is logical, but as I have stated, I am not a Biologist and do not know exactly how the stains work relative to the cell information you requested.
I hope this helps and again apologize for the delay.
Best of Luck,
Bob