Update impute.R to 1) clarify impute_gmt_titers parameters and 2) enforce reproducibility

[TKMarkCheng]

1. update documentation to clarify required result for impute_gmt_titers(result,titer) is from titertools::gmt() rather than circular dependency.

2. enforces a random seed variable with a default to help with reproducibility of imputation results for less experienced users

[shwilks]

Thanks for these changes, the changes to the documentation look good, thank you for doing that. For the addition of random seeds within functions, typically the method in R packages is simply to use set.seed() at the top of your script, outside of the functions to ensure reproducibility, or if you really want, to set it in your script directly before the function is called.

Unless there is a good reason to set the seed within a function this is generally avoided, since it creates some confusion for the user about what the seed is at any given point. If you absolutely do want to change the seed within a function for some reason, it should also be restored again once the function (e.g. runif()) has been called, otherwise it will mess up the seed for other things the user runs as part of the script in ways they might not be expecting.

Happy to hear if there is a use case where you couldn't simply do the following for reproducibility though

set.seed(100)
impute_gmt_titers(...)

[TKMarkCheng]

Thanks Sam! I've added the within function set.seed() as a failsafe/reproducibility guarantee for new users who are unaware of set.seed(), or are running them in a more iterative way e.g. in a .rmd since set.seed() calls the seed at run time rather than stores it as a global variable.

I agree with your point that it can create some confusion by changing the random seed mid-script and may produce more issues down the line. Agree with keeping it simple. Happy for just the comment amendment to be pulled.