Which is a great shame – as I passionately believe that it’s a powerful, yet simple concept. It’s moving from measuring one variable at a time, to multiple. And getting a better, more accurate picture off the back of it.
Sounds great in theory… except that it can become overwhelming in practice.
I remember trying to create my first full factorial design for media optimization, with 3 factors each being measured at 3 levels. That added up to 27 runs.
3^(3) = 27 runs
But then I got ambitious and increased the number of factors to seven. All of a sudden, the number of runs jumped from 27 to an unmanageable 2187.
3^(7) = 2187 runs
And that was just the beginning…
Switching between software makes the DOE process harder
I was lucky enough to learn about DOE early on in my career. But even with all the right theory, my first hands-on experience with planning DOE was challenging.
It amazed me that there wasn’t a single piece of software that covered the whole DOE process from start to finish. Switching from a spreadsheet, to DOE design software, back to a spreadsheet to define concentration points, create my design, and generate an execution plan disrupted my thought process.
And the learning curve was steep. I could see that every DOE design solution out there was powerful – no one can deny that they can do some pretty epic stuff. However, they aren’t always intuitive and if you don’t have a stats background, they can be super confusing to use.
I joined Synthace’s lab team after hearing about the changes in store for their DOE product. I knew instantly that these changes would make the power of DOE more accessible to biologists – and it got me excited.
Having the option to choose a templated DOE design within the Synthace workflow builder, and automatically map it on to an experiment design, would be a game changer. It would mean that you wouldn’t need to know a lot about DOE to get instant value from it.
Even DOE software super-users would have something to gain, as you would be able to upload your own DOE designs.
And the prospect of doing a streamlined DOE under one roof – the experiment planning, execution or automation, and data analysis stages, all in one place – was the cherry on top of the cake.
You can create D-optimal, space-filling and full-factorial designs with support for “hard-to-change” factors, “quasi-replicate” factors, and multi-level categories within Synthace’s workflow builder.
Can I still design my DOE with another software if I want to?
If you’re a super-user of JMP, DesignExpert or MODDE and have a particularly complex design in mind, you can still run your DOE from the Synthace DOE workflow. Simply generate your DOE design as you normally would, then upload a .csv file to Synthace’s workflow.
What types of factors can I use on the Synthace DOE workflow?
Our workflow supports the following types of factors:
Factor – a multi-level factor that will be included in the workflow and design analysis
Derived factor – when the levels of a factor are generated by another factor, either by a numerical formula or categorical mapping
Mutual exclusion – when only one factor in a group will be active at any given time
At the click of the button, you can also further qualify your factors as:
Quasi-replicate factors – factors that are almost exact replicates of another factor, but with slightly different properties
Hard-to-change (HTC) factors – factors that are difficult to change, i.e., due to time, cost, and more commonly, temperature. This button will cause the design to add replication to your HTC factors, helping you discern the effect of the factor on your response.
There’s also the sampling section – which allows you to mark whether your factors only apply to spacefill designs, where you may wish to make some factors only selected either at:
Discrete intervals, i.e., used only at the levels that you define
Range, i.e., used at as many levels as there are runs in the DOE design
Let me take you through a DOE design that you can do for media optimization – a great example of where DOE can get you fantastic results.
Say you wanted to ascertain the optimum levels of media constituents for mammalian cell growth. Your factors could be epidermal growth factor and hepatocyte growth factor concentration – and you could measure your response by optical density.
1. Define your factors and levels
In the “DOE” tab of the Synthace workflow builder, you can define:
The constituents of the growth media that you’d like to include as design factors
The concentration / volume levels for each design factor
2. Choose and optimize your DOE design
Next, you select an appropriate DOE design, which will be automatically generated and calculated for you.
At this point, you’ll be presented with your design’s predicted “statistical power” value – the probability that your chosen design will generate statistically significant results.
Based on the value you’re presented with, you can decide whether you want to boost your “statistical power” by increasing the number of design runs, or choosing a different DOE design entirely.
3. Check everything will run as planned
Before executing your experiment manually or on an automated liquid handler, you can simulate your DOE design and preview the entire experiment in silico – even from the comfort of your own home.
DOE is an iterative process. After executing your first DOE experiment, you can come back to your original workflow and seamlessly modify your factors, levels, and design – and generate a new simulation within seconds.
Got another burning question that I haven't covered?