You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
## The maximal number of cores to use for FastQC, STAR, Salmon and DRIMSeq. ## Note that the actual number of cores available to Snakemake is determined by ## the --cores argument when it is invoked.ncores: 1
and then in some rules you have:
threads:
config["ncores"]
I got confused a couple of moths ago when I didn't know snakemake as I do know. As you say, with --cores you specify the maximal number of cores used by snakemake, so when your are running the rule star_SE with snakemake --cores 12 and you specify in the config file ncores: 2, you'll be running 6 instances of STAR in parallel, but each instance will be using 2 cores, you won't be running STAR in one file with 12 cores (one instance). I didn't get that from the config file. I'm saying this because it's important to know how many input files you have (approximately) to set ncores properly and, I think, many times is better to set threads independently in each rule because of this and also, because different software has different performance when running with multiple cores in parallel. My intention isn't that you change the behavior of this pipeline, It is rather to try to avoid confusing (unintentionally) some users (like me). Maybe you can add some information in the wiki or in the config file about this.
The text was updated successfully, but these errors were encountered:
In the configuration file,
config.yaml
, saysand then in some rules you have:
I got confused a couple of moths ago when I didn't know
snakemake
as I do know. As you say, with--cores
you specify the maximal number of cores used bysnakemake
, so when your are running the rulestar_SE
withsnakemake --cores 12
and you specify in the config filencores: 2
, you'll be running 6 instances ofSTAR
in parallel, but each instance will be using 2 cores, you won't be runningSTAR
in one file with 12 cores (one instance). I didn't get that from the config file. I'm saying this because it's important to know how many input files you have (approximately) to setncores
properly and, I think, many times is better to setthreads
independently in each rule because of this and also, because different software has different performance when running with multiple cores in parallel. My intention isn't that you change the behavior of this pipeline, It is rather to try to avoid confusing (unintentionally) some users (like me). Maybe you can add some information in thewiki
or in theconfig
file about this.The text was updated successfully, but these errors were encountered: