FISH-Quant: the sequel

As promised, I started using FISH-Quant to analyze my FISH images.

I must say that I enjoy using FQ much better than the previous program that was developed by one of my lab members. I find FQ more intuitive, more informative, and with a better range of parameters and tools for analysis.

In essence, the program goes through several steps:

In the first step, one must determine basic parameters (such as NA, pixel size, ex/em wavelengths). After further defining folders and loading the image file one needs to define an outline of the cell.

Determining cell outline, nucleus & transcription sites is fast & easy.

Determining cell outline, nucleus & transcription sites is fast & easy.

This is already very helpful – in the previous program, I either analyzed an entire image, or had to draw the outline (mask) in ImageJ, then use a third program to select spots only in the mask area. I could only use one mask per analysis. with FQ, I can use multiple outlines per image, and I can get automated detection of the nucleus (based on a DAPI image) and automated identification of transcription sites inside the nucleus.

At this point I should mentioned that I came across some bugs – which initiated a very productive work relations with Florian Mueller – the developer of FQ.

The next step is the filtering of the image. I’m not exactly sure how its done mathematically. From the little I understand, the program first blurs the image. This simulates the background, which is then subtracted from the original image. The program then enhances small features that remained.

Filtering: on the left - the "blurred" image. In the middle - after subtracting the background. On right - the final filtered image.

Filtering: on the left – the “blurred” image. In the middle – after subtracting the background. On right – the final filtered image.

Then comes a pre-detection step. In this step, a range of thresholds is tested for detecting the spots. After it is done, one chooses the best threshold to give “true spots”, based on a quality score of the spots.

Pre-detection of the spots. Several parameters to play with.

Pre-detection of the spots. Several parameters to play with.

Here, Florian implemented a new feature: in the original (published) program, this step was performed on the entire cell. Since I’m currently interested only in cytoplasmic mRNAs, and the nucleus gave lots of background, Florian added a feature that enables me to analyze only the cytoplasm or only nucleus (or both). This was very helpful.

The next step is fitting the parameters from the pre-detection to the original image and you get results.

Results! green circles - "good" spots. Blue circles: "bad" spots. the histograms on the lower left help determine which threshold to use

Results! green circles – “good” spots. Blue circles: “bad” spots. the histograms on the lower left help determine which threshold to use

Thresholding of the spots can be done on several paramaters: XY and Z sizes and amplitude of the signal. The histograms below help determine the best thresholds. The program also shows composite images of maximum projection in XY and XZ of all spots.

One can also take a closer look using the spot inspector.

Inspecting the spots is fun!

Inspecting the spots is fun!

This window allows inspection of each and every spot. At my request, Florian added another feature, of enabling the index number of each spot to appear next to the spot. Unfortunately, there is still a little bug there, which doesn’t really interfere:

Numbers, numbers all around me!

Numbers, numbers all around me!

A main advantage of FQ, is the ability to do batch analysis of images (from the filtering step onward. One still needs to draw outlines in each image individually. Here again Florian helped by adding a quick save button). Batch analysis allows better thresholding (the statistics are better since one can look at all spots in all images in one histogram).

In summary – this has been a good experience so far. I am still trying to find the best threshold for my FISH images but I’m learning a lot in the process.

The major advantage of FQ is actually its ability to analyze transcription sites more precisely than other programs (I elaborated on that in the previous post). I haven’t even started using FQ for that, but I will, once my yeast project will run smoothly…

 

The original paper for further reading on how this program works:
ResearchBlogging.org Mueller, F., Senecal, A., Tantale, K., Marie-Nelly, H., Ly, N., Collin, O., Basyuk, E., Bertrand, E., Darzacq, X., & Zimmer, C. (2013). FISH-quant: automatic counting of transcripts in 3D FISH images Nature Methods, 10 (4), 277-278 DOI: 10.1038/nmeth.2406

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s