Optimal designs are most useful for screening, exploration and optimization. However, they can rely on you having some prior knowledge of your biological system such that you can set the upper and lower levels of your factors. Setting your factor levels in the wrong place could lead to a lot of 0’s in your data, which could lead to poor models in your analysis.
In this tutorial you will learn:
How to select and calculate an optimal design.
Defining an optimal design
Once two or more factors have been defined, click on “Calculate Design” in the DOE design window to choose a DOE design. To learn how to define factors and their levels, click here.
Click on the Optimal Design tab and Synthace will start calculating a design automatically. Once the design calculation has completed you will see design diagnostics in the Diagnostics panel. To learn more about design diagnostics, click here.
Control runs can be specified in the Controls panel by clicking on Add and manually filling in the table with factor levels. To learn more about defining control runs, click here.
The calculated factor levels for each run can be found in the Design panel. You can also decide how to group runs together based on factors and their levels and allocate them to different simulations for execution purposes here. To learn more about grouping runs to different simulations, click here.
Configuring your optimal design.
Specify the number of runs you want to investigate in the final design in the settings panel. If this is increased to a number greater than the number of wells in the output plate selected, Synthace will create new plates to satisfy the Design Run number.
Note: changing the value in this field will trigger a recalculation of your design.
Seed is used to seed the calculation of your optimal design. This value ensures the calculation is deterministic for the specific random seed, so designs can be recalculated with the same result. Any integer value can be provided here and the design will be recalculated. Changing this value can change the distribution of factors and levels for the runs in your experimental space.
Note: changing the value in this field will trigger a recalculation of your design.
Select the type of model you intend to fit to your data during response analysis. The choice of model influences the calculation of your optimal design. Given a specific model choice the combination of factors and their levels will be calculated to give the best opportunity to fit that type of model during data analysis.
The options available are for main effects, up to two factor interactions and up to quadratic effects models.
If you only have numerical factors with two levels each then you will only be offered models for main effects and up to two factor interactions.
If you have numerical factors that have three or more levels you will be given the option to fit up to quadratic models too.
Note: changing the value in this field will trigger a recalculation of your design.
Specify how many times you would like to replicate your entire optimal design in the Replicates input in the settings panel.
Note: changing the value in this field will trigger a recalculation of your design.
If you have specified more than one replicate in the replicates panel you can also define how you want the runs in those replicates to be ordered. By default the replicate order will be to repeat the same run order in each of the replicates of the design.
You can also choose to keep each replicate run next to one another, randomize all runs across all replicates or randomize the runs within each replicate group independently.
Note: changing the replicate ordering option will not recalculate your design, but the order of the runs will be updated in the design matrix at the bottom of the page. To learn more about replicating run order, click here.
Once your design has been calculated - you can click Simulate With Design. To learn more about simulating workflows with a DOE design, click here.
Well done on making it to the end of this tutorial.
To learn about other design types, click here.
To learn more about replicating run order, click here.
To learn how to assess the quality of your design with design diagnostics, click here.
To learn more about how to define control runs, click here.
To learn more about grouping runs to different simulations, click here.