Design of Experiment (DoE) with R at SIVB 2023

In 2023, I presented at the Society for In Vitro Biology's (SIVB) Design of Experiment Workshop. This workshop aimed to introduce the in vitro biology community to the invaluable tool of Design of Experiments (DoE). My presentation specifically demonstrated that the open-source programming language R can be used for this purpose instead of expensive commercial software.

About the presentation

My presentation at the Society for In Vitro Biology's (SIVB) Design of Experiment Workshop, was preceded by Dr. Randall Niedz, a DoE authority in our professional community who has published numerous papers demonstrating applications of DoE in tissue culture. A recording of the workshop is available on YouTube.

In the DoE presentation, I showcased the advantages of using a scripting language like R for Design of Experiments (DoE) and guided participants through the code for designing and analyzing experiments. The presentation featured key optimization techniques, such as central composite design and mixture-process design, as well as my preferred screening method, the fractional factorial design. A complete version of the PowerPoint is available below.

Learning DoE is like picking money off of the floor.

Following my presentation, we discussed the need for free, entry-level DoE software, as modern commercial options cost over $1000 and offer only a short trial period. This is inadequate for tissue culture experiments, which often require more than a month for incubation between design and analysis. Given the interest, I proposed creating a Shiny app using R for this purpose. A first draft of the app is available on this webpage.

I learned DoE using StatEase, which I highly recommend for practitioners not interested in learning R. StatEase Design Expert offers numerous design options, comprehensive help pages, an intuitive UI, and high-quality graphics. However, it costs over $1000 per year. I was fortunate my boss allowed me to purchase this software after discovering its use in a paper on culture media optimization. Unfortunately, many people who could benefit from this skill may not have the same opportunity. Therefore, a freely available tool could be valuable before justifying the need for expensive commercial software.

While this free DoE software is not intended to replace or compete with STATease or JMP, it can be valuable for learning experimental design principles. The Shiny app is currently in its first draft and is being actively tested by the professional and academic community. It was my first large-scale Shiny app, developed before the availability of LLMs for coding assistance. If there is significant interest and we identify ways to improve the UI for a more intuitive design experience, I plan to create an enhanced version using the {golem} framework. This modular approach would be better suited for collaborative, open-source development.

For those interested in learning how to use R for experimental design, I will publish several blog posts on this page. These posts will include examples from tissue culture experiments I conducted as the Research Director at Tissue-Grown Corporation. In addition to experimental design insights, they will feature exclusive tissue culture research not published elsewhere.