It is often useful to divide a set of data into distinct groups, or classes. Phaedra supports this notion on the subwell level, through the concept of subwell classification. A common scenario is cell classification, where cells are labeled to distinguish living cells from dead cells, fluorescence-expressing cells versus non-expressing cells, etc.
Before classification can be performed, a classification feature must first be defined. This is a feature whose value is the class that has been assigned to each subwell item.
For example, you can define a classification feature called ‘Cell State’ with two classes: Living and Dead.
For more information about setting up a classification feature, see Editing the Protocol Class.
With manual classification, classes are assigned to cells by selecting groups of cells on an image, in a table or in a plot.
Once a selection has been made, a class can be assigned to the selection by clicking on the classification button:
- Classification feature: this box shows the available classification features. While this is exceptional, some protocols have multiple classification features. Select the one you want to use here.
- Available classes: this table lists the available classes for the selected classification feature. Each class has a name, a color, a symbol and an optional description. Furthermore, the number of items (e.g. cells) that match the class is shown as a number and a percentage of the entire well.
- Affected items: this table lists the items that will be affected by the classification. To change this set, close the dialog, make another selection and click the classification button again.
- Save or close: when you are done making changes, you can either click Save to keep the changes or Close to discard the changes.
After confirming the classification, the items will be updated with their new class value. This may trigger a recalculation, for example when calculated well features have been defined that use a classification feature in their formula.
Manual classification is not always feasible. If you need to perform classification on a large number of wells, you may consider using automated classification. This requires the use of a workflow. See Using Workflows for more information about running a workflow, or contact a Phaedra administrator to help you with setting up a new workflow.
Of course, automated classification only works if the classification criteria can be applied by the computer itself, without user intervention or visual inspection. A typical example of automated classification is applying a fixed threshold on a subwell feature that expresses marker intensity.