Nested Factor DoEs for BioassaysHello Everyone,
Had a customer involved in bioassay development ask how easy it is to create nested DoE's - in these cases they may have only one attempt to use a certain reagent lot or cell preparation with a certain analyst, they are never repeated/shared amongst different analysts.
They currently use the Minitab Gauge R+R (Nested) tool for this - can we offer an easier/better way to approach this?
Re: Nested Factor DoEs for BioassaysHi Ben,
I think the MSA design capability in JMP 17 offers the flexibility needed: https://www.jmp.com/support/help/en/17.2/index.shtml#page/jmp/example-of-msa-design.shtml#
However it is unusual for a user to switch from Minitab to JMP unless we can steer them toward an important need they have that they can't solve in Minitab. Use the answer you give them regarding our capabilities to design nested experiments to earn the right to ask why they are considering switching from Minitab to JMP. What is motivating them to consider changing?
Re: Nested Factor DoEs for BioassaysHi Ben,
I've seen something like this before in biotech companies working on cancer cell therapeutics. Often there is a cell donor effect that is treated as a random block effect. A single random blocking effect is not hard to put into the custom designer. But things do get a little hairy when the donor effect is nested within another blocking effect, like analyst or day. (Or vice versa, day is nested within donor).
When you have two nested random effects, you can hack the custom designer to produce the appropriate design. The trick is to create two dummy factors (Call them D1 and D2), and call one hard to change, the other very hard to change. After producing the design table, you can rename the whole plot column "Day", and the subplot column "Donor". Then delete the two dummy columns. It's not a pretty hack, but it works well. And this approach is more flexible than using a gauge R&R study framework. Let me know if this is unclear and I can record a short video.
By the way, there's an excellent Discovery talk about modeling donor effects appropriately, highly recommended: https://community.jmp.com/t5/Discovery-Summit-Americas-2021/Biopharmaceuticals-New-Frontier-Modeling-Patient-Donor/ta-p/398736
