Closed pcm32 closed 5 months ago
Hi @pcm32,
The way I coded it is that we run multiple genesets in parallel already using numba, and it does not matter what the input is (AnnData or dataframe). Hope this is helpful!
Thanks. Is there a way to control then the number of concurrent processes that would run should you give it various gene sets at the same time in the network? I'm assuming this is how you achieve the parallelisation, correct? Perhaps through a numba max number of processes?
By default it uses all available cores, but you can control the number of threads by changing this global parameter:
import numba as nb
nb.set_num_threads(n=32)
# Then run decoupler functions here
Running AUCell for multiple gene sets can be time consuming, so ideally one would like to run multiple genesets in parallel. Given the current implementation of run_aucell, is this feasible and thread-safe? Or will it only work if the input is not an AnnData object that gets modified at the end? Thanks!