Entering edit mode
7 weeks ago
tekica
•
0
Hi, I need to run a pseudo-alignment with Kallisto on scRNA-seq data. I have around 300GB of fastq files that need to be aligned. I have access to the HPC cluster, but I am new to it so I don't know how to properly set up sbatch parameters such as --cpus-per-task, --ntasks and --mem or any other that I should use. Here are the specifics of the nodes:
- 12 compute nodes equipped with two 32-core AMD EPYC 7452 processors (64 virtual cores) and 256GB of RAM.
- 12 compute nodes equipped with two 8-core Intel Xeon E5-2630 processors (16 virtual cores) and 64GB of RAM.
I would really appreciate it if you could give me advice on how to set up these parameters (and suggestions for any additional) so that I don't overload the node, but still get the job running. Thank you in advance!!!
Just do --cpus-per-task of 16, --mem of 16G, and --ntasks of 1. Request 12 hours of time.
You won't need more than that to efficiently process your single-cell RNA-seq data (actually, it's already way overkill), and those fit within the specification of all compute nodes listed.
When setting up slurm jobs, just give generous ballpark estimates and go with that; if the job fails, just re-run it under different settings.