0

I'm bootstrapping a glmmTMB using glmmboot package (version 0.5.1) in R. This useful package is implemented with the future package in order to parallelize processes. When I run the command it gives me this warning:

UNRELIABLE VALUE: Future (‘future_lapply-1’) unexpectedly generated random numbers without specifying argument 'future.seed'. There is a risk that those random numbers are not statistically sound and the overall results might be invalid. To fix this, specify 'future.seed=TRUE'. This ensures that proper, parallel-safe random numbers are produced via the L'Ecuyer-CMRG method. To disable this check, use 'future.seed=NULL', or set option 'future.rng.onMisuse' to "ignore"

I know that other users has reported this issue: https://github.com/satijalab/seurat/issues/3622 and the answer were that it weren't a problem because in that case as it wasn't using random numbers. Moreover, this warning should have been fixed in previous versions of the package (I'm using version 1.21.0).

My concern is about if in my case this is a problem or if I could ignore the warnings.

StupidWolf
  • 45,075
  • 17
  • 40
  • 72
Luca Di Nuzzo
  • 1
  • It is impossible to say for "your case". But they are likely correct in their assumption that "it does not matter". The reason basically is that it should run `RNGkind("L'Ecuyer-CMRG")` first, to use a parallel-stable RNG machine, if the numbers need to be "truly" random. Dependng on what the random numbers are used for (here I am assuming they split datasets on each cluster) it may or may not matter. I would not expect a significant problem here either way, due to the small amount of random numbers used for bootstrapping. – Oliver Mar 31 '21 at 10:13
  • You could likely run the command yourself before bootstrapping if you are concerned. See `?RNGkind` for more information on random numbers in R. – Oliver Mar 31 '21 at 10:15

0 Answers0