diff --git a/internal_use/docs/source/AdvancedSegmentation/BBBC022_AnalysisExercise.md b/internal_use/docs/source/AdvancedSegmentation/BBBC022_AnalysisExercise.md index ff7be64cb..76862a352 100644 --- a/internal_use/docs/source/AdvancedSegmentation/BBBC022_AnalysisExercise.md +++ b/internal_use/docs/source/AdvancedSegmentation/BBBC022_AnalysisExercise.md @@ -74,7 +74,7 @@ a similar introductory exercise. ## **Input images and configure metadata** -### 1. **Load images and metadata** +### **1. Load images and metadata** - Start CellProfiler by double-clicking the desktop icon CellProfiler icon @@ -86,7 +86,7 @@ a similar introductory exercise. images (with a file extension of ‘.npy’) are included in this data set. -### 2. **Import metadata from the CSV** +### **2. Import metadata from the CSV** So that we can explore what cells treated with different drugs look like later in the exercise, we must add this information into CellProfiler from the CSV. Provided with this exercise is a CSV called ‘20585_AE.csv’ detailing drug treatment info for each image. @@ -113,7 +113,7 @@ info for each image. - Image_Metadata_PlateID (from the spreadsheet) is matched to Plate (extracted from the folder name by the second extraction step) - Image_Metadata_CPD_WELL_POSITION (from the spreadsheet) is matched to Well (extracted from the file name by the first extraction step) -### 3. **Examine the channel mappings in NamesAndTypes (optional)** +### **3. Examine the channel mappings in NamesAndTypes (optional)** The channel mapping here is a bit more complicated than anything we've worked with before- we have a single set of illumination correction images that map to each and every well and site. We can use the metadata we extracted in the last module to make that association possible. @@ -154,7 +154,7 @@ The channel mapping here is a bit more complicated than anything we've worked wi ``` ## **Illumination correction** -### 4. **Examine the output of the CorrectIlluminationApply module (optional)** +### **4. Examine the output of the CorrectIlluminationApply module (optional)** Since microscope objectives don't typically have a completely uniform illumination pattern, applying an illumination correction function can help make segmentation @@ -183,9 +183,9 @@ to the top of the field of view to see the greatest effect. *Figure 4: Application of the illumination correction functions.* ``` -## **Segmenet Nuclei, Cells and Cytoplasm** +## **Segment Nuclei, Cells and Cytoplasm** -### 5. **IdentifyPrimaryObjects- Nuclei** +### **5. IdentifyPrimaryObjects- Nuclei** Next we'll take a first pass at identifying nuclei and cells in our initial image. @@ -202,7 +202,7 @@ Next we'll take a first pass at identifying nuclei and cells in our initial imag the parameters for robustness later, however, the identification should be good but doesn’t need to be perfect before you move on. -### 6. **IdentifySecondaryObjects- Cells** +### **6. IdentifySecondaryObjects- Cells** - **After** the IdentifyPrimaryObjects module but **before** the EnhanceOrSuppressFeatures module, add an IdentifySecondaryObjects @@ -215,7 +215,7 @@ Next we'll take a first pass at identifying nuclei and cells in our initial imag you feel you’re ready to test them on another image; they need not be perfect before you move on. -### 7. **Test the robustness of your segmentation parameters across multiple compounds** +### **7. Test the robustness of your segmentation parameters across multiple compounds** It's (relatively!) easy to come up with a good set of segmentation parameters for a single image or a set of similar images; this data set however contains images from cells treated with many different classes of drugs, many of which have very different phenotypes. It's valuable to learn how to create a set of parameters that can segment cells that display a variety of morphologies since you may come across a similar problem in your own experiments! @@ -259,7 +259,7 @@ It's (relatively!) easy to come up with a good set of segmentation parameters fo \- In IdentifyPrimaryObjects, adjusting the declumping settings (make sure to turn 'Use advanced settings?' on) will probably be necessary for a robust segmentation \- In IdentifySecondaryObjects, you will want to test the effects of using the various methods for identifying secondary objects (Propagation, Watershed-Image, Distance-N, etc) and, if using Propagation, the regularization factor. -### 8. **IdentifyTertiaryObjects- Cytoplasm** +### **8. IdentifyTertiaryObjects- Cytoplasm** - **After** the IdentifySecondaryObjects module but **before** the EnhanceOrSuppressFeatures module, add an IdentifyTertiaryObjects @@ -269,7 +269,7 @@ It's (relatively!) easy to come up with a good set of segmentation parameters fo ## **Segment Nucleoli inside the Nuclei** -### 9. **Examine the steps used to segment the Nucleoli** +### **9. Examine the steps used to segment the Nucleoli** - The next 3 modules have to do with the creation of the Nucleoli objects. Look at the output from each to see how the image is @@ -308,7 +308,7 @@ It's (relatively!) easy to come up with a good set of segmentation parameters fo ## **Segment the Mitochondria inside the Cytoplasm** -### 10. **Mask the Mito image by the Cytoplasm object** +### **10. Mask the Mito image by the Cytoplasm object** Now that you’ve seen an example of how to segment an organelle, you will do so for Mitochondria in the following steps. @@ -335,7 +335,7 @@ will do so for Mitochondria in the following steps. ``` -### 12. **IdentifyPrimaryObjects- Mitochondria** +### **11. IdentifyPrimaryObjects- Mitochondria** - **After** your MaskImage module but **before** the RelateObjects modules, add an IdentifyPrimary Objects module to identify @@ -350,7 +350,7 @@ will do so for Mitochondria in the following steps. ## **Perform Measurements** -### 13. **Add measurement modules to your pipeline** +### **12. Add measurement modules to your pipeline** - **After** your segmentation of the mitochondria but **before** the RelateObjects modules, add as many object measurement modules as you @@ -426,7 +426,7 @@ only be able to examine one object at a time in CellProfiler Analyst.* ## **Relate Nucleoli and Mitochondria to their respective nuclei/cells** -### 14. Examine the settings of RelateObjects +### **13. Examine the settings of RelateObjects** - **After** your Measurement and **before** your Export modules you should find two RelateObjects modules. One relates Nucleoli to @@ -439,7 +439,7 @@ only be able to examine one object at a time in CellProfiler Analyst.* ## **Perform the analysis on ALL the images** -### 15. **Run the pipeline (optional)** +### **14. Run the pipeline (optional)** - If you have time and/or if you’d like to play with the data in CellProfiler Analyst later, exit test mode, close the eyes next to diff --git a/internal_use/docs/source/BeginnerSegmentation/CPbeginner_Segmentation.md b/internal_use/docs/source/BeginnerSegmentation/CPbeginner_Segmentation.md index 8df4a3283..b5014917c 100644 --- a/internal_use/docs/source/BeginnerSegmentation/CPbeginner_Segmentation.md +++ b/internal_use/docs/source/BeginnerSegmentation/CPbeginner_Segmentation.md @@ -58,7 +58,7 @@ Read through the steps below and follow instructions where stated. Steps where y :width: 700 :align: center - Figure 2: **Main CellProfiler window**. To load images, drag and drop images into the right area. To load a pipeline (.ccpipe or .ccproj files), drag and drop the pipeline file into the left area. + Figure 2: **Main CellProfiler window**. To load images, drag and drop images into the right area. To load a pipeline (.cppipe or .cpproj files), drag and drop the pipeline file into the left area. ``` - Drag and drop the `‘segmentation_start.cppipe’` file into the `‘Analysis modules’` pane on the left. @@ -94,9 +94,9 @@ Read through the steps below and follow instructions where stated. Steps where y > **TIP** you can manually adjust brightness and contrast in the image display by right-clicking on it and going to `'Adjust Contrast'` ------------------------------------------------------------------------------------------------------------------------- -### 3. [OPTIONAL STEP] Set up the input modules +### **3. [OPTIONAL STEP] Set up the input modules** - > *We suggest you skip this step for now, it will not affect the rest of the pipeline, as these modules have been properly set up in the starting pipeline (`segmentation_start.cpipe`).* + > *We suggest you skip this step for now, it will not affect the rest of the pipeline, as these modules have been properly set up in the starting pipeline (`segmentation_start.cppipe`).* > *At the end of this tutorial you will find instructions on how to set up the input modules* diff --git a/internal_use/docs/source/QualityControl/BBBC022_QualityControlExercise.md b/internal_use/docs/source/QualityControl/BBBC022_QualityControlExercise.md index e36b2d239..c52adf3fe 100644 --- a/internal_use/docs/source/QualityControl/BBBC022_QualityControlExercise.md +++ b/internal_use/docs/source/QualityControl/BBBC022_QualityControlExercise.md @@ -46,7 +46,7 @@ tools in CellProfiler and CellProfiler Analyst, preferably after completing the Translocation tutorial or a similar introductory exercise. -## 1. **Start the provided QC pipeline on the BBBC022 dataset** +## **1. Start the provided QC pipeline on the BBBC022 dataset** In order to do quality control, we need to first measure the images in many ways. This will allow us to do machine learning to use the measurements to identify the @@ -73,7 +73,7 @@ good images from the bad. three files should be created- a .db database file, a .properties text file, and a .workspace text file. -## 2. **Examine the QC pipeline (~15 minutes)** +## **2. Examine the QC pipeline (~15 minutes)** - While the pipeline is running, take some time to look over the pipeline and make sure you understand the various parts. You will @@ -150,7 +150,7 @@ good images from the bad. table, etc). As we are not identifying any objects, we don’t need to worry about these. -## 3. **Open the CellProfiler Analyst workspace and determine reasonable parameter cutoffs (~20 minutes)** +## **3. Open the CellProfiler Analyst workspace and determine reasonable parameter cutoffs (~20 minutes)** In the first step of the quality control pipeline, we'll look at graphs of how various measurements are distributed in the population. This allows us to get @@ -211,7 +211,7 @@ in later steps. different/smaller subset or delete it altogether by using the Gate Inspector (‘gate’-> ‘MANAGE GATES’). -## 4. **Optional — use the PlateViewer tool to check for other features to gate on (~10 minutes)** +## **4. Optional — use the PlateViewer tool to check for other features to gate on (~10 minutes)** If you want to see if you can find additional features that might distinguish good images from bad images, feel free to explore the feature set more thoroughly. @@ -239,7 +239,7 @@ images from bad images, feel free to explore the feature set more thoroughly. ImageNumber on the X-axis as in the Workspace plots, but feel free to play around). If you find a gate that seems logical to make, proceed as in Step 3. -## 5. **Create filters based on the cutoffs you’ve determined (~10 minutes)** +## **5. Create filters based on the cutoffs you’ve determined (~10 minutes)** Now that we've created gates around our poor quality images, we need to convert them into filters so that we can access them in the Classifier tool. This @@ -317,7 +317,7 @@ presented in case you cannot or will not edit the properties file. *Figure 3: Creating filters inside CPA* ``` -## 6. **Create classifier rules to distinguish good from bad images (~30 minutes)** +## **6. Create classifier rules to distinguish good from bad images (~30 minutes)** Creating gates based on the 12 measurements we graphed has helped us identify some low quality images so far, but we are not utilizing the rest of our @@ -359,7 +359,7 @@ low quality images. - Save your training set for future reference if desired, then close CellProfiler Analyst. -## 7. **Add quality control steps to an analysis pipeline (~15 minutes)** +## **7. Add quality control steps to an analysis pipeline (~15 minutes)** If you have time, you can add the list of rules you identified in your machine learning classifier to the CellProfiler pipeline that corresponds to this data diff --git a/internal_use/docs/source/Translocation/Translocation.md b/internal_use/docs/source/Translocation/Translocation.md index 54ecaa642..2c51ece34 100644 --- a/internal_use/docs/source/Translocation/Translocation.md +++ b/internal_use/docs/source/Translocation/Translocation.md @@ -76,7 +76,7 @@ treated and untreated cells. ## Exercise I: Using the CellProfiler software to identify features and obtain measurements from cellular images. -### 1. **Starting CellProfiler and configuring the input data for analysis** +### **1. Starting CellProfiler and configuring the input data for analysis** - Start CellProfiler by double-clicking the desktop icon @@ -154,7 +154,7 @@ treated and untreated cells. - Click the “Update” button below the divider to display a table that shows each channel pair matched up for the 26 wells in the assay. -### 2. **Identifying the nuclei as the “primary objects” that you will analyze** +### **2. Identifying the nuclei as the “primary objects” that you will analyze** Now that the module inputs and outputs are set up, in your module, the remaining settings need to be adjusted to best detect the nuclei. The @@ -229,7 +229,7 @@ The result may look like Figure 3. *Figure 3: A zoomed-in view of the display window for IdentifyPrimaryObjects* ``` -### 3. **Improve identification of primary objects** +### **3. Improve identification of primary objects** In this instance, in Figure 3, you can see that the outlines capture too much of the background around the nuclei. This means that the default @@ -288,7 +288,7 @@ them as foreground pixels, leaving only the lowest intensity pixels as background. The identified outlines should now better match the actual nuclei boundaries. -### 4. **Identifying the cell body as a “secondary object” that you will analyze** +### **4. Identifying the cell body as a “secondary object” that you will analyze** Now that you have confirmed, by eye, that the settings we provided you in this exercise do allow for identification and segmentation of the @@ -354,7 +354,7 @@ of pixels without regard to the underlying fluorescence. appears underneath to 10 pixels. - Click the button to see the result from your new settings. -### 5. **Identifying the cytoplasm as a “tertiary object”** +### **5. Identifying the cytoplasm as a “tertiary object”** Once we have identified the nucleus and the cell body, these two objects can be used to define the cell cytoplasm as the region outside the @@ -383,7 +383,7 @@ objects, effectively identifying the cytoplasm. *Figure 5: Example module display window for IdentifyTertiaryObjects* ``` -### 6. **Measuring the cells’ characteristics (i.e. the “object features”)** +### **6. Measuring the cells’ characteristics (i.e. the “object features”)** Now that the objects have been identified using settings that have been optimized for the phenotypes of interest, the next step is to make @@ -470,7 +470,7 @@ measurements. - Select “MeanIntensity” from the “Measurement” drop-down list. Then select “rawGFP” from the “Image” drop-down that appears. -### 7. **Creating an image with your cell and nuclear outlines on it (optional)** +### **7. Creating an image with your cell and nuclear outlines on it (optional)** It’s often nice to create an image showing the segmentation of your objects so that you can refer back to it later; in addition to the @@ -536,7 +536,7 @@ The **SaveImages** module can be used to either save images generated in any ste “\_Overlay” is appropriate. - All the other settings may be left at their default values. -### 8. **Exporting the measurements to a database** +### **8. Exporting the measurements to a database** Since we will be using the data visualization and machine learning tools in CellProfiler Analyst, the measurements will need to be saved to a @@ -575,7 +575,7 @@ Analyst to access them. Command-click (Mac). Leave the rest of the settings at the default values. -### 9. **Using the optimized pipeline to automatically analyze all images generated by the screening experiment** +### **9. Using the optimized pipeline to automatically analyze all images generated by the screening experiment** At this point, the settings you have entered were chosen for you because those settings specifically, when used with these images, result in an @@ -633,7 +633,7 @@ have extracted from the cells. - If you move the database file, you'll need to edit the properties file to point to the new database location. -### 1. **Visualizing the measurements in a 96-well plate layout view** +### **1. Visualizing the measurements in a 96-well plate layout view** CPA has several tools available for displaying the data for exploration. If your data came from a multi-well plate, such as the 96-well plate for @@ -701,7 +701,7 @@ visualization tools available is the plate layout format. - Do not close the Plate Viewer tool, as you will be referring to it later in the exercise. -### 2. **Using the Classifier function of CPA to distinguish the cells’ FOXO1A-GFP subcellular localization phenotypes** +### **2. Using the Classifier function of CPA to distinguish the cells’ FOXO1A-GFP subcellular localization phenotypes** CellProfiler Analyst contains a machine-learning classification tool, which will allow you to distinguish different phenotypes automatically. @@ -769,7 +769,7 @@ sorting. **Bottom:** *Examples of positive cells (left) and negative cell - We refer to this set of positive and negative cells you have assembled as the “training set.” -### 3. **Reviewing the rules that CPA established (based on your training set) to classify positive and negative cells** +### **3. Reviewing the rules that CPA established (based on your training set) to classify positive and negative cells** The classification rules you will examine below are CPA’s way of defining the measurements (and the cutoff values the measurements need @@ -794,7 +794,7 @@ phenotypes. measurement one that you would expect to be the most significant one to use in distinguishing the phenotypes? -### 4. **Reviewing the accuracy of the classification with the confusion matrix** +### **4. Reviewing the accuracy of the classification with the confusion matrix** Once you have trained a classifier, you can test the ability of the of the classification rules to predict which class each cell in your @@ -834,7 +834,7 @@ belong to. While the cells in this simple example were able to be predicted perfectly, that is rare in real data.* ``` -### 5. **Refining the training set by sorting more “unclassified” cells into the “positive” and “negative” bins** +### **5. Refining the training set by sorting more “unclassified” cells into the “positive” and “negative” bins** At this point, it is important to keep in mind that the CPA Classifier tool will pick whichever measurement is most significant in making its @@ -937,7 +937,7 @@ of the desired phenotype; (iii) Correct misclassifications, or sort into appropriate bins; (iv) Go back to the first step and repeat, until the classifier displays the desired level of accuracy. -### 6. **Classifying all cells in the experiment** +### **6. Classifying all cells in the experiment** Once the classifier is of the desired accuracy, it is ready to be applied to the complete image data set. @@ -972,7 +972,7 @@ You can also save your training set and/or classifer model for future reference or to make changes later; do so by going to *File > Save Training Set* or *File > Save Classifier Model* -### 7. **Saving the scores to the measurement database for visualization** +### **7. Saving the scores to the measurement database for visualization** Now that we have successfully scored our experiment, we will save the scores back to the measurement database, so that they can be visualized @@ -1008,7 +1008,7 @@ using CPA’s tools. of “*Image_Metadata_Dose*”? (2) What does this correspondence (or lack thereof) tell you about the classifier? -### 8. **Plotting the scoring results, to estimate the lowest dose necessary to induce FOX1O-GFP translocation** +### **8. Plotting the scoring results, to estimate the lowest dose necessary to induce FOX1O-GFP translocation** You can use additional data tools in CPA to visualize your data in other ways. In this case, we will use a scatter plot to plot a dose-response @@ -1033,7 +1033,7 @@ cells with GFP in the nucleus) increases with Wortmannin dose. lowest dose that produces an enrichment score similar to that of the maximum dose? -### 9. **Exporting your classifier for use in a CellProfiler pipeline** +### **9. Exporting your classifier for use in a CellProfiler pipeline** Head back to the Classifier tool. The *File-->Save Classifier Model* menu option can be used to export a .model file which stores the trained