The Experimental Design step allows you to specify or edit information about your experimental design, such as the primary factors studied, the subjects (for within-subject designs), blocks (for blocked designs), or any other variables that you might want to investigate. This information is used throughout the workflow to propose specific alignment setups, image display layouts, appropriate statistical tests and data analysis tools.
Confused about what experimental design, factors, factor levels, subjects and blocks are? Read our Experimental design guide for a brief introduction on theses concepts.
The first time you work through the Experimental design step, a wizard helps you define one of the more common experimental designs.
If your experiment design does not match one of the designs in the wizard, just Skip this step. You can define experimental design variables further on in the process.
First choose whether your experiment includes one (top row) or two (bottom row) primary factors.
Then click on the thumbnail of the applicable design:
- Pick the option …with subjects if the experiment collects multiple measurements from each subject. For instance, you may have taken a sample from each subject before, during and after a treatment.
- Pick the option …with blocking if the experimental units (sample run on a gel, or a combination of gel and dye for DIGE experiments) have been matched into homogeneous blocks before they were randomly assigned to a treatment group. For instance, you may have run your 2D gel experiment on different days, each time with several treatments handled in parallel. Since you anticipate day-to-day variability, you want to analyze the data in such a way that each day is treated as a matched set, called block.
- Pick the default option if you do not have a within-subject or blocked design.
Click OK to move to the next step.
In the second screen of the wizard, name your factors by modifying the default names.
For each factor:
- Name each factor level by double clicking on the default name.
- Add factor levels by clicking on the green plus sign.
- Remove factor levels by clicking on the red cross of the factor level.
- Change the color of a factor level by double-clicking on its color box.
- Optionally, in the String in file column, add a string that identifies the factor level in the file name. This automatically assigns images with this string in their file name to the factor level.
- For each subject, specify the Level for subject. This means that you specify within which level of the second factor the subject is nested.
Any of the above changes will automatically be reflected in the experimental design matrix at the bottom of the display. If all factor levels are identified by strings in the file names, the images are automatically inserted in the matrix. If this is not possible, you can assign the images to their factor levels in the next step.
When you are finished, click the Create button.
Add or delete factors/variables
If you have defined factors with the wizard, you will see them listed here as headers in the Experimental design table at the top left of the screen. If you have skipped the wizard, you will only see the list of your images in an empty table.
You can define (additional) factors, variables or information fields that you might want to investigate during the analysis. Click on the green plus sign to create a new factor. Please note that the software uses “factor” as a general term for primary factors, variables and other information fields.
The software will display a Properties window where you can set the properties of your new factor:
- Name of the factor.
- Type of the factor. The available types are:
- Factor – Use this type for the primary factors that you want to investigate.
- Blocking – Use this type to identify blocks (groups of experimental units that are formed to be as homogeneous as possible with respect to the block characteristics) in a blocked design. The levels for such factor correspond to the different blocks.
- Subject – Use this type to identify the subjects in a within-subject design, i.e., when you have collected multiple measurements for each subject. The levels for such factor correspond to the different subjects.
- Information – Use this type to identify other experimental variables that you might want to look at during the analysis (e.g. technician that ran the gels, …)
- Levels: Once the factor levels have been defined, click on the corresponding color box to change their color.
Note that the Properties window is also accessible when clicking on the Properties icon that appears behind the name of a selected factor.
Remove factors by clicking on the red cross next to the factor name.
When a new factor is created, the thumbnails of all validated images in the project will appear in the Images area in the middle of the screen. To create a new factor level, you can select images (they will be highlighted in green) and drag their thumbnails into the new level… box in the Levels area at the bottom of the screen.
Once you dragged images into the new level… box, the default level name is highlighted so you can edit it.
Select images based on strings in their names by entering the string in the Select box at the top of the screen. You can also select images manually, either by clicking on their thumbnail in the Images area or on their name in the Experimental design table. Hold the Shift or Ctrl key to select multiple images.
Edit factor levels
To edit factor levels, the corresponding factor must be selected.
- Remove images from a factor level by dragging their thumbnails back to the Images area or to another level.
- Rename a factor level by clicking on its name in the Levels area.
- Delete a factor level by clicking its red cross in the Levels area.
- Change the color of a factor level by clicking the Properties icon for the selected factor, and then clicking the color box for the factor level.
Run the wizard again
You can create a new design by clicking the Wizard button.
The Experimental design matrix at the top right of the screen summarizes the main factors in your experiment: up to two primary factors, as well as the subject or blocking factor.
Layout and colors of the experimental design matrix
If you do not have an Experimental design matrix displayed on your screen, or if you want to change its layout, you need to adjust the layout settings for your factors: In the Experimental design table, click on the Change display icon that can be found below each factor. Choose Display vertically, Display horizontally or Not displayed and see the change in the matrix. Click the Reset display to wizard choice icon in the Experimental design table to reset the layout to the design that was created using the wizard.
The color coding in the Experimental design matrix corresponds to the color coding for the selected factor. To select a factor, just click on its name in the Experimental design table.
Consistency of the experimental design matrix
Use the Experimental design matrix to verify that your design is consistent and balanced, meaning you have the same or similar number of biological replicates for each treatment.
Designs with a subject or blocking factor must be fully balanced: no missing values are allowed. If, for a design where samples were taken from each subject before, during and after a treatment, the after observation is missing for a given subject, the experiment cannot be analyzed as a within-subject design (using the appropriate repeated measures ANOVA). Instead, the subject factor will be ignored, and the experiment will be analyzed as a between-subject design (using ordinary ANOVA).
Note that some specialized statistical packages make corrections to manage unbalanced designs. After alignment, detection and normalization, you can therefore export your quantification data for analysis with such third-party software.
Experimental design used for the statistical analysis
The default analysis in the Results step will use the experimental design that was selected and specified using the wizard.
If you skipped the wizard and defined your design manually, the default analysis in the Results step will use the design visualized in the Experimental design matrix once you proceed to the next step in the workflow.