diff --git a/AdvancedSegmentation/BBBC022_Analysis_Final.cppipe b/AdvancedSegmentation/BBBC022_Analysis_Final.cppipe index a088c4b4..c52702ae 100644 --- a/AdvancedSegmentation/BBBC022_Analysis_Final.cppipe +++ b/AdvancedSegmentation/BBBC022_Analysis_Final.cppipe @@ -355,7 +355,7 @@ MeasureColocalization:[module_num:14|svn_version:'Unknown'|variable_revision_num Calculate the Rank Weighted Colocalization coefficients?:Yes Calculate the Overlap coefficients?:Yes Calculate the Manders coefficients using Costes auto threshold?:Yes - Method for Costes thresholding:Fast + Method for Costes thresholding:Faster MeasureObjectIntensity:[module_num:15|svn_version:'Unknown'|variable_revision_number:4|show_window:False|notes:[]|batch_state:array([], dtype=uint8)|enabled:True|wants_pause:False] Select images to measure:ER, Mito, Hoechst, Ph_golgi, Syto diff --git a/AdvancedSegmentation/README.md b/AdvancedSegmentation/README.md index a0bc53e0..f70c3b1f 100644 --- a/AdvancedSegmentation/README.md +++ b/AdvancedSegmentation/README.md @@ -8,4 +8,4 @@ They are fields of U2OS cells (10 sites imaged in 5 channels), and represent 5 mock treated wells from a single 384 well plate experiment. The download bundle for this tutorial lives in a public google drive folder: -https://drive.google.com/drive/folders/1t6h2e4ifjRISQaXvFg-Mr9mOfc0vET9C?usp=sharing +https://drive.google.com/file/d/18YB8RhfFNKZmYFBd2I0mujJNkcxxEuIZ/view?usp=sharing diff --git a/BeginnerSegmentation/Archive_EN.zip b/BeginnerSegmentation/Archive_EN.zip index 2d25d5df..a6f8aef9 100644 Binary files a/BeginnerSegmentation/Archive_EN.zip and b/BeginnerSegmentation/Archive_EN.zip differ diff --git a/BeginnerSegmentation/Archive_ES.zip b/BeginnerSegmentation/Archive_ES.zip index 7cb15de0..304fdcca 100644 Binary files a/BeginnerSegmentation/Archive_ES.zip and b/BeginnerSegmentation/Archive_ES.zip differ diff --git a/BeginnerSegmentation/Archive_PT.zip b/BeginnerSegmentation/Archive_PT.zip index 95a8de17..c466d6c8 100644 Binary files a/BeginnerSegmentation/Archive_PT.zip and b/BeginnerSegmentation/Archive_PT.zip differ diff --git a/BeginnerSegmentation/bonus_materials.zip b/BeginnerSegmentation/bonus_materials.zip index f0451ee4..c67f0fde 100644 Binary files a/BeginnerSegmentation/bonus_materials.zip and b/BeginnerSegmentation/bonus_materials.zip differ diff --git a/BeginnerSegmentation/bonus_materials/beginner_segmentation_bonus_exercises.pdf b/BeginnerSegmentation/bonus_materials/beginner_segmentation_bonus_exercises.pdf index e258d711..13f27d69 100644 Binary files a/BeginnerSegmentation/bonus_materials/beginner_segmentation_bonus_exercises.pdf and b/BeginnerSegmentation/bonus_materials/beginner_segmentation_bonus_exercises.pdf differ diff --git a/BeginnerSegmentation/segmentation_final.cppipe b/BeginnerSegmentation/segmentation_final.cppipe index a72a9299..b7f2aaeb 100644 --- a/BeginnerSegmentation/segmentation_final.cppipe +++ b/BeginnerSegmentation/segmentation_final.cppipe @@ -239,7 +239,7 @@ MeasureColocalization:[module_num:14|svn_version:'Unknown'|variable_revision_num Calculate the Rank Weighted Colocalization coefficients?:Yes Calculate the Overlap coefficients?:Yes Calculate the Manders coefficients using Costes auto threshold?:No - Method for Costes thresholding:Fast + Method for Costes thresholding:Faster RelateObjects:[module_num:15|svn_version:'Unknown'|variable_revision_number:5|show_window:True|notes:['This module relates a larger "Parent object" to another smaller "Child object/s". In this case, this allows us to identify which Nucleoli belong to which Nuclei.', 'The output is a new object called NucleoliChildObjects that has all of the nucleoli belonging to a single nucleus ']|batch_state:array([], dtype=uint8)|enabled:True|wants_pause:False] Parent objects:Nuclei diff --git a/Instructions_for_adding_tutorials.md b/Instructions_for_adding_tutorials.md index ea934b8c..316de906 100644 --- a/Instructions_for_adding_tutorials.md +++ b/Instructions_for_adding_tutorials.md @@ -3,15 +3,15 @@ 1. Gather all the necessary files in a folder. Give the folder a descriptive name with no spaces. Note that files > 25 MB (and especially >100 MB) may not be uploadable to GitHub. If your tutorial requires large files, please do the following: 1. Add the files to a folder with a descriptive name (no spaces) 2. Zip the folder (on mac, right click > Compress...) - 3. Add the .zip file to google drive in the `Outreach` > `Workshops_Webinars_Tutorials` folder on the shared g drive + 3. Add the .zip file to google drive in the `Outreach` > `Workshops_Webinars_Tutorials`> `GithubLinkedTutorialFiles` folder on the shared g drive 4. Make sure the folder is public and copy the link to share the folder to use for the download link later. 2. Make a PR adding the folder of materials to the tutorial repo. In addition to the materials, please also 1) write a short description of the tutorial to add to the readme and - 2) create a gif (tips on making one [here](https://docs.google.com/document/d/1G5odCSbX1fW8vReKH5ubgvWc1sIGquLisUe7t5pR7d8/edit#heading=h.kjidkmwv6jq3)) and add that gif to to the `assets` > `img` folder. + 2) create a gif (tips on making one [here](https://docs.google.com/document/d/1G5odCSbX1fW8vReKH5ubgvWc1sIGquLisUe7t5pR7d8/edit#heading=h.kjidkmwv6jq3); in 2024 on Mac, Beth likes Gifski) and add that gif to to the `assets` > `img` folder. 3) add the tutorial to the `internal_use/docs/source/index.rst` file, and the index file of whatever language that tutorial is in (ie if English, `internal_use/docs/source/toc_en/index.rst`) -3. Update the readme with the description and gif. If there is an associated blogpost or video, please link that using the appropriate badge. If the tutorial is particularly geared toward beginners, add the beginner badge. If the tutorial is advanced such that a user should definitely do a beginner tutorial first, use the advanced badge. +3. Update the top-level readme with the description and gif. If there is an associated blogpost or video, please link that using the appropriate badge. If the tutorial is particularly geared toward beginners, add the beginner badge. If the tutorial is advanced such that a user should definitely do a beginner tutorial first, use the advanced badge. ## Here's an example of what the markdown looks like for the Translocation tutorial: ``` diff --git a/README.md b/README.md index 89d13f9b..d8d2ae08 100644 --- a/README.md +++ b/README.md @@ -40,11 +40,11 @@ This exercise will guide you through setting segmentation parameters that will b example segmentation -Download materials +Download materials En español: -Materiales para descargar +Materiales para descargar ## Translocation diff --git a/internal_use/docs/source/BeginnerSegmentation/BeginnerSegmentationPlugins.md b/internal_use/docs/source/BeginnerSegmentation/BeginnerSegmentationPlugins.md index dd9e015e..a84c77f8 100644 --- a/internal_use/docs/source/BeginnerSegmentation/BeginnerSegmentationPlugins.md +++ b/internal_use/docs/source/BeginnerSegmentation/BeginnerSegmentationPlugins.md @@ -100,7 +100,7 @@ Cell masks were made by navigating into the image folder and then running the co 1. Open up a clean copy of CellProfiler (or run File -> New Project) and drag `bonus_1_import_masks.cppipe` into the pipeline panel. 1. Drag the `images_Illum-corrected` subfolder from the main Beginner Segmentation exercise and also the `cellpose_masks_cells` subfolder from this exercise. *Do not drag in the `cellpose_masks_nuclei` folder.* 1. Put CellProfiler into TestMode , open the eye icon next to OverlayOutlines, and then hit 3 times to create a classical segmentation and compare it with the Cellpose-generated version. You can check the settings in the OverlayOutlines module to see which outline color corresponds to which segmentation in the output. -1. Optionally, open the Workspace Viewer using to create easily on-the-fly customizable overlays. +1. Optionally, open the Workspace Viewer using to create easily on-the-fly customizable overlays. 1. Hit and repeat a couple of times to examine the segmentations on more images. ```{admonition} Question for you @@ -120,7 +120,7 @@ Visualization of images and objects in the Workspace Viewer ### Bonus-to-the-Bonus - add the provided nuclear segmentations as well 1. Drag and drop the `cellpose_masks_nuclei` folder in the Images module to add the nuclear masks to the image list. -1. Go to the NamesAndTypes module and configure it so both kinds of masks can be loaded (hint: there's a duplicate button!) +1. Go to the NamesAndTypes module and configure it so both kinds of masks can be loaded (hint: there's a duplicate button you can use to make a second copy of a channel!) - You'll probably have to modify the existing settings for loading the cell masks by changing its rule or adding a second rule. Do you understand how/why? ## **Exercise 2: Running ilastik locally or from a Docker container** @@ -242,6 +242,28 @@ The output of the RunCellpose module 4. As before, using OverlayOutlines and/or the WorkspaceViewer, evaluate segmentation on a few images. Where is CellProfiler doing better, and where is Cellpose doing better? 5. Use the button to learn more about the different parameters you can pass to Cellpose (we don't offer all of them, but many!) - how does tweaking these affect your output? How about changing the model you're using, and/or the image you're segmenting? +### Bonus-to-the-Bonus 1 - test on a more varied data set + +The data we gave you in this exercise was all very clean, and all from the negative control wells of this experiment. The [folder linked here](https://drive.google.com/file/d/10kZMhr2bVR14vMP7kG5iZGujjUI3Nn3Y/view?usp=sharing) contains images with a broader variety of phenotypes - try replacing the images in the `Images` module with this folder instead and see how well the two kinds of segmentation work in different conditions! + +```{figure} ../AdvancedSegmentation/TutorialImages/Fig2.png +:width: 700 +:align: center +A broader range of phenotypes can be harder to consistently account for in conducting segmentation. +``` + +```{tip} +You’re encouraged to explore the performance in some random images on your own (you can do this from the `Test` menu with the `Random Image Set` option), but if you find yourself consistently ending up with images that look similar you can try examining images from the following list of wells- A08, A12, B12, B18, C7, D6, D19, D22, E3. You can pick these with `Test` -> `Choose Image Set` +``` + +### Bonus-to-the-Bonus 2 - improve Cellpose segmentation + +Based on your evaluations above, can you identify some places where additional training might help fix some issues in the Cellpose model? If you happen to have Cellpose on your computer (or want to try installing it - see Exercise 1 for a link to the official install instructions), open Cellpose with some of the `Ch1` images and see if you can train a model that will perform better. + +### Bonus-to-the-Bonus 3 - test your classical segmentation skills + +The classical segmentation parameters provided here are based on expert tuning from a very experienced CellProfiler user - how close can you get? Add another IdentifyPrimaryObjects module to the pipeline, tune it yourself, and then look in the workspace viewer to see how it performs, especially in the more diverse image set from Bonus-to-the-Bonus 1. Compare your parameters to the expert ones - how do they differ? Do you understand why some of these settings may have been chosen? + ## What next? Want to know more about CellProfliler plugins and modules? 1. Read the [CellProfiler-plugins paper](https://pubmed.ncbi.nlm.nih.gov/37690102/) 2. Watch this [video](https://www.youtube.com/watch?v=fgF_YueM1b8) to learn how to write a module