set_array_job: 'Set up array job on farm'

In yangjl/huskeR: Genomic and Genetic analysis Pipeline on HPC

Description

Farm is a computer cluster running slurm system. Note, bigmem mem=8000 per cpu, hi med low mem=25000 per cpu, serial mem=1500 per cpu.

Usage

set_array_job(shid = "largedata/GenSel/CL_test.sh",
  shcode = "sh largedata/myscript.sh", arrayjobs = "1-700",
  wd = NULL, jobid = "myjob", email = NULL, runinfo = c(TRUE,
  "bigmemh", "1", "10G", "8:00:00"))

get_runinfo(runinfo)

Arguments

shid	Relative or absolute path and file name of your shell code, i.e. CL_test.sh.
shcode	The commands inside your sh file.
arrayjobs	A character specify the number of array you try to run, i.e. 1-100.
wd	Working directory, default=NULL. It will use your current directory.
jobid	The job name show up in your sq NAME column.
email	Your email address that farm will email to once the job was done or failed.
runinfo	Parameters specify the array job partition information. A vector of c(TRUE, "bigmemh", "1"): 1) run or not, 2) -p partition name, and 3) –cpus.

Value

return a shell file.

Examples

for(i in 1:10){
    shid <- paste0("slurm-script/run_", i, ".sh")
    command <- paste0("bedtools getfasta -name -tab -fi roast.chrom.", i, ".msa.in")
    cat(command, file=shid, sep="\n", append=FALSE)
}
shcode <- paste("module load bismark/0.14.3", "sh slurm-script/run_$SLURM_ARRAY_TASK_ID.sh", sep="\n")

set_array_job(shid="slurm-script/run.sh", shcode=shcode,
              arrayjobs="1-10", wd=NULL, jobid="myjob", email=NULL,
              run = c(TRUE, "bigmemh", "8196", "1"))

yangjl/huskeR documentation built on Sept. 2, 2021, 5:38 a.m.