A component of CellProfiler cell image analysis software
The WormToolbox is a toolbox for high-throughput screening of image-based Caenorhabditis elegans phenotypes. The image analysis algorithms measure morphological phenotypes in individual worms and are effective for a variety of assays and imaging systems. WormToolbox is available through CellProfiler and enables objective scoring of whole-worm high-throughput image-based assays of C. elegans for the study of diverse biological pathways that are relevant to human disease [Nature Methods paper].
The main analysis pipeline, Pipeline 1, shown in blue, is fully automated, and consists of the following steps:
If the provided default worm model does not fit the input data (e.g. due to a different image resolution or different worm strain), a new worm model can be created using Pipeline 2 (pink) to (g) manually select non-touching worms making up a training set for a new model built by Pipeline 3.
For low-throughput experiments, it is possible to manually curate faulty segmentation results using Pipeline 4 (orange) by (h) manual editing of the output from step d of pipeline 1, and (i) manual flipping of digitally straightened worms so that they all align in with heads or tails up. Measurements, such as intensity of stain, fat stain distribution, fat region size, worm width etc. may also be extracted from individual worms after manual correction. All measurements are exported to a database for further exploration and phenotyping.
Note that CellProfiler 2.1 should be used.
Pipeline 1: Identify and collect measurements from individual worms and sub-regions
CellProfiler 2.1 pipeline: Pipeline1_UntangleWormsExtractMeasurements.cppipe
Worm model: DefaultWormModel
Download this video: Pipeline1.mp4
This is the main pipeline for delineating worms (also those in clusters) and extracting different kinds of shape and intensity measurements. It requires a worm model as input. This model could either be the DefaultWormModel.xml provided above, or a new worm model created by pipelines 2 and 3 above. This pipeline is fully automated, does not need any user input (once optimized), and can be run on a large number of images organizing output based on information extracted from the input file names. Example image data is provided below.
Steps of the pipeline:
Pipeline 2: Find, select, and save individual worms
CellProfiler 2.1 pipeline: Pipeline2_SelectSingleWorms.cppipe
Download this video: Pipeline2.mp4
This pipeline may be omitted if a previously created model (e.g. DefaultWormModel.xml provided with Pipeline 1) fits the new data. Steps A-C are the same as in Pipeline 1; the same pre-processing and foreground/background segmentation should be applied both when creating the worm model and when running the analysis as changes to the pre-processing may affect the appearance of the width of the worms, in turn influencing the model. This pipeline requires manual interaction selecting representative non-touching worms.
Steps of the pipeline:
NOTE: The video incorrectly states that WormObjects should be used as input for the ConvertObjectsToImage module. The correct input for that module is SelectedSingleWorms.
Pipeline 3: Create a new worm model from individual worms
CellProfiler 2.1 pipeline: Pipeline3_TrainModel.cppipe
Download this video: Pipeline3.mp4
This module takes the manually selected training worms from Pipeline 2 as input. It is recommended to visually browse through the binary input images before running this pipeline to make sure that they represent single worms. The pipeline is fully automated, and the resulting model will be called MyWormModel.xml.
Pipeline 4: Untangle, correct, and straighten worm clusters
CellProfiler 2.1 pipeline: Pipeline4_ManuallyCorrectStraighten.cppipe
Worm model: DefaultWormModel.xml
Download this video: Pipeline4.mp4
This pipeline loads the untangling results from Pipeline 1 and allows the user to edit any errors in the untangling prior to straightening the worms for visualization and extraction of measurements. A worm model (DefaultWormModel.xml or a new model created by Pipelines 2 and 3) is needed as input for the straightening step.
Previously published example pipelines using the WormToolbox
As presented in Wählby et al. 2012, these additional pipelines include illumination correction and feature extraction from fluorescence microscopy data.