Extending an experiment

[llrs]

Sometimes people sequence 50 samples and then wants to add 25 more samples to that cohort. Explain some problems on the vignette. Provide a specific function.

• Check which are old and which new
• Check if any variable matches that (warn, that results will be confounded) (perhaps check that in the normal functions too)
• Suggest which old samples to add to avoid batch effects.

[llrs]

This other package can be relevant: https://github.com/hoehleatsu/socialroulette/

[dwuab]

I'm also interested in this issue too, but couldn't find anything in the literature. I'm not even sure if "extending an experiment" is the right terminology to use here, since a google search of the phrase would yield many irrelevant results.

[llrs]

Hi Wu, thanks for commenting showing your interest! Given that someone else is interested I might dedicate some time to provide a solution to this. Probably extending experiment is not accurate ( I am not native speaker, so there might be better expressions for it) but I haven't come up with or found a better name for it.

On the CRAN task View of Experimental Design there might be something, perhaps adaptive design or stage-design are better, or we might something relevant on the references provided.

[llrs]

It might be interesting to use Minirand for this (check this thread in twitter), instead of provide my own solution (there is some code at one of the references' supplementary information).

[llrs]

Hi @dwuab, I've added some functions in the devel version (soon to be released). There are a follow_up and follow_up2 functions to help with this.

I am still unsure about the name, let me know if these functions help you. Note that I also added a new function check_data to provide some feedback about the data.

What I haven't added yet is suggesting some samples to avoid a batch effect.