CiliaQ Output Parameters
Here we provide an overview about the output parameters that you generate with CiliaQ, how to retrieve them from the output file, how they are calculated, and how they can be interpreted.
Tutorial written by Lilian Fu and Jan N. Hansen in December 2021.
After processing your images with the CiliaQ plug-in, multiple files for each analyzed image will be generated. This tutorial will focus on the parameters, which are ouput in the three .txt files generated by CiliaQ:
-
CQ.txt: This file is most commonly used. It contains the output for all the commonly used parameters plus an additional header that describes general information, such as the saving date and the settings that were used for processing.
-
CQl.txt: This file can be used for a post-hoc analysis of the ciliary shape and subciliary protein localization. It provides additional, very specific information that is not saved in the CQ.txt file, such as coordinates of all ciliary points with the geometric curvature of a cilium. It might also be of interest if you need the intensity profile in a format that can easily be used for further analysis and read by programs such as R or Matlab. While the CQ.txt file provides one row per cilium, this CQl.txt file provides one row per ciliary point. Points from the same cilium do all provide the same ID. The Point-ID sorts the ciliary points from base to tip, the arc length indicates the calibrated position of the point on the cilium, from base to tip. Further analyzing intensity profiles may be useful to check where exactly your protein of interest is localized in the cilium.
Excerpt of a CQl.txt file showing Point-IDs and corresponding parameters for cilium #4
- CQs.txt: This file is a short version of the CQ.txt file, containing the same information but without the header. You can use this to more easily compile information without disruption by the header.
What to do with my .txt files?
A) One option is to simply open the appropriate .txt file and copy-pasting everything into another program, such as excel. When using the CQ.txt file it might look something like this:
In the Results part, columns show different parameters, rows show different cilia. The “ID” is unique for each cilium – it allows to connect the data for the same cilium in the CQ.txt file and the CQl.txt file. You can also find the IDs in output images, which allows you to look up where the information for the cilium comes from.
Images show the identified cilia, labelled with their ID. Left image: “SKL_3D.png” file, right image: “CQ_RP.tif” file
You may then want to copy the values for a column of your parameter of interest into a program like GraphPad Prism or visualize them directly using Excel.
B) If copy-pasting each individual file seems like a bit of a hassle you can also feed the .txt file into an R script instead (-> see R Scripts`) to help better organize, adjust and visualize the data.
Which parameters are relevant to you will depend on your biological question but you can find an explanation on all the CiliaQ output parameters here. Note, that all parameters depend on the segmentation method that was selected in CiliaQ Preparator, but this can be more evident for some parameters than others. It is therefore essential to use the same imaging and processing settings when comparing the absolute values from different samples.
Another thing to keep in mind is the effect of the Gaussian blur selected in the skeletonization settings in CiliaQ. It affects only skeleton-based parameters (e.g. “centerline” parameters and intensity profiles). Generally, the blur helps eliminate side branches to create more defined cilia but a higher blur will also lead to shorter cilia. You might need to optimize these settings to find those that are suitable for your data. The figure below depicts how different Gaussian blur settings impact cilia parameters.
Source: Hansen, et al. CiliaQ: a simple, open-source software for automated quantification of ciliary morphology and fluorescence in 2D, 3D, and 4D images. Eur. Phys. J. E_ 44, 18 (2021). doi: 10.1140/epje/s10189-021-00031-y. Licensed under CC BY 4.0.
- x-, y-, z-center [μm]: These represent the coordinates for each cilium. They can be used to investigate the distribution of the cilia, e.g. in the context of an organ or tissue to determine if the cilia show a localization preference. As an example, in the figure below we determined the ciliary density at different brain regions in a zebrafish embryo.
Source: Hansen, et al. CiliaQ: a simple, open-source software for automated quantification of ciliary morphology and fluorescence in 2D, 3D, and 4D images. Eur. Phys. J. E_ 44, 18 (2021). doi: 10.1140/epje/s10189-021-00031-y. Licensed under CC BY 4.0.
-
Surface [μm2], Volume [voxel, μm3]: Note that these parameters are highly dependent on the settings used for imaging as well as processing, as the threshold directly affects the values.
-
Shape complexity index: This parameter describes how complex the structure of each cilium is – a cilium that is perfectly spherical will have a shape complexity index of 1. The higher the index the more complex the cilium is, meaning the surface has a lot of “bumps” or the cilium is branched or may be very long and have a lot of bends. High resolution images that are able to show branched cilia might be required. This parameter is also highly dependent on the threshold and image noise.
Cilium on the right has a higher **shape complexity index** than the one pictured on the left.
-
Sphere radius [μm]: Used to calculate the shape complexity index and may only be relevant for a very specific post-hoc analysis involving additional computations.
-
Number of found skeletons, branches: These parameters can be used to check the quality of your analysis. For example, having a lot of tree branches can be an indicator that you need to adjust your settings to improve segmentation and reduce the background noise. However, also in well analyzed data sets with good settings, you will find side branches. So a lot of tree branches may not directly be a criterion to exclude a cilium (-> see also this Q&A in the CiliaQ wiki).
-
Cilia length [μm]: Measures the shortest path based on the skeleton of each cilium. The skeletonization function reduces each ciliary object to the finest possible line by removing pixels to create the skeleton for each cilium. Note that not every cilium may yield a skeleton: spherical cilia will not reveal a skeleton because in the skeletonization process they are reduced to a point, which has no length. Thus, these spherical cilia will show up blank in the output table. You may edit this in R or Excel by replacing N/A with 0 and adding a small arbitrary value to the length of each cilium if desired. In our analysis, we usually convert NAs to 0s and add the length of one pixel to all length values to also consider spherical cilia in length estimates.
Example of skeletonization process for a cilium and a perfectly round object
-
Tree length [μm]: You can use the this parameter to measure the length of a branched cilium since the cilia length parameter only takes the shortest distance into account and disregards side branches.
-
Orientation vector x, y, z [μm]: Determined by the first and last point of the skeleton. The first point is marked in the skeleton output files with a cyan point and is considered as the base of the cilium. The orientation vector has a direction but without a basal stain it is difficult to determine the correct orientation of the cilium. If no basal stain was added and indicated in CiliaQ, the base detected by CiliaQ may not be the actual base of the cilium. You may still draw conclusions from this parameter based on the orientation of the cilia relative to the x, y, or z axis of the image, e.g., by determining the angle between orientation vector and the y axis of the image.
- Cilia bending index: This index describes how much a cilium is bent – the higher the index the more it is bent. Note that this parameter does not correspond to the mathematical definition of curvature but still it relates to the curvature.
These parameters are measured in the channels specified by you, depending on how you labelled your protein of interest.
-
Minimum, maximum, SD of intensity: These parameters can indicate whether the intensity of the signal differs across the cilium. The more heterogenous the intensity level in the cilium is, the larger the determined SD will be and the more the minimum and maximum will differ. For example, if your protein of interest is only localized at the tip of the cilium the SD of intensity will be high and differences between minimum and maximum will be large.
-
Profile A, B (Fluorescence intensity profile): The values for this parameter are obtained by interpolating the pixels along the centerline of the cilium. This parameter is dependent on the orientation of the cilium. When looking at the ..._SKL_3D.png output image, the basal part of each cilium as determined by CiliaQ is represented by the cyan dot (see image below). Without a basal stain it is not always clear which tip of the cilium is really the base. However, you can align the cilia by a custom post-hoc analysis in R, for example if you know your protein of interest shows a localization preference.
-
Average intensity: This parameter considers all the voxels contained in the entire ciliary volume to determine the average intensity of the signal. It is, therefore, influenced by how broad the mask is.
-
Average intensity on centerline: This parameter is less influenced by the broadness of the mask than the average intensity since it only takes the centerline of each cilium into consideration to calculate the average intensity of the signal.
-
Integrated fluorescence intensity on centerline: This parameter can be used if you want to compare the total fluorescence intensity between cilia, i.e. the amount of protein of interest contained in different cilia. Notably, while the average intensity considers the length of the cilium and is independent of the length, in the integrated intensity the total protein amount is determined and thus, longer cilia may have higher values than shorter cilia, even if the average intensity is smaller.
-
Average of 10% highest pixels: This parameter evaluates the ciliary voxels sorted by intensity and looks at the 10% of the ciliary voxels with highest intensities in each cilium. You may want to use this parameter if you have a broad mask and a very small signal in order to circumvent dilution of the signal or if you want to determine the average intensity of a signal that is only present in a part of the cilium.
-
Colocalized volume A, B [μm3, % of total volume]: This parameter counts the voxels that are assigned a value in channel A or B, meaning voxels that are not = 0. To use this as a meaningful parameter, these channels need to be segmented. Note that in segmented channels, you should not measure the absolute intensity parameters (e.g. average, SD, …) simultaneously because it may be biased by zero values of pixels removed in the segmentation process.
-
Colocalized compared to background volume [μm3, % of total volume]: This parameter reflects how much higher the measured signal is compared to the background. The background is defined as the signal in the cell soma. For this, the image is divided into 25 equal cuboids. The voxels with the highest 10% signal of each cuboid, that are not part of a cilium are used to calculate the mean + SD for the entire image. This is then defined as the background. Note, that if you have highly fluorescent cell somas or other structures in the image this will largely affect the determination of the threshold. This can be particularly problematic when these structures do not occur in all images, because an independent threshold is determined for every image. Under such circumstances, it may be better to also segment the intensity channel with a fixed threshold using CiliaQ Preparator and use the following parameter.