Hi everyone, I would appreciate your help on the ...
# help
t
Hi everyone, I would appreciate your help on the following (I am sure it has been asked before). I am trying to set up nf-core pipelines on a local lab server at UCSF on an NFS-mounted storage system (/mnt/nfs/...) My goal is to have a fully self-contained, optimized setup including: 1. PIPELINE ASSETS - nf-core/rnaseq - nf-core/rnavar 2. CONDA ENVIRONMENTS - Pre-built conda envs for each tool (STAR, GATK, Salmon, etc.) - Shared across pipelines and users - Stored in a central NFS location 3. REFERENCE FILES - iGenomes GRCh38 (NCBI) — already downloaded locally - STAR index - GATK bundle (dbSNP138, Mills indels) - VEP cache - snpEff cache - dbNSFP 4. QUESTIONS: What is the recommended way to: a) Pre-build ALL conda environments before running b) Set a shared conda cache directory accessible to all users? My current command looks like: nextflow run nf-core/rnavar \ -r 1.1.0 \ -profile conda \ --input samples.csv \ --outdir results \ --genome GRCh38 \ --igenomes_base /mnt/nfs/.../iGenomes \ --star_index /mnt/nfs/.../STARIndex \ --fasta /mnt/nfs/.../genome.fa \ --gtf /mnt/nfs/.../genes.gtf \ --dbsnp /mnt/nfs/.../dbsnp138.vcf.gz \ --known_indels /mnt/nfs/.../Mills.vcf.gz \ -resume Any advice on institutional configs or best practices would be greatly appreciated! Thanks in advance!
p
There are many ways to do this, but a couple of quick notes: • I'd strongly advise using Singularity/Apptainer instead of conda if possible. Better reproducibility and less error prone. No shared builds needed then, just a shared singulairity cache dir (`$NXF_SINGULARITY_CACHEDIR` and `$NXF_SINGULARITY_LIBRARYDIR` were created for exactly this) • See https://nf-co.re/configs for a list of institutional configs, might find some good stuff in there. Users can then run
nextflow run -profile my_institute
and have a bunch of parameters pre-set (such a igenomes_base etc) For more complete preconfiguration, one approach I've seen folks do is use the environment module system to pre-configure entire offline pipelines + assets. See https://docs.uppmax.uu.se/software/nextflow/ Whether you really need this or whether it's overkill is your call.
t
Thank you very much. Phil ! Very helpful indeed !
šŸ‘šŸ» 1