ewels / clusterflow

A pipelining tool to automate and standardise bioinformatics analyses on cluster environments.
https://ewels.github.io/clusterflow/
GNU General Public License v3.0
97 stars 27 forks source link

Software installation #121

Closed YuntaoTan closed 5 years ago

YuntaoTan commented 5 years ago

Should I install bowtie manually or CF will install automatically? Or other external softwares, such as BWA? if this can be solved will be great!

ewels commented 5 years ago

Hi @YuntaoTan,

Cluster Flow was designed to work with environment modules on HPC systems. These are centrally installed software packages that can be dynamically loaded with commands such as module load bowtie.

Each CF tool wrapper has a list of environment modules that it tries to load. For example, the bowtie2 module tries to load bowtie2 and samtools: https://github.com/ewels/ClusterFlow/blob/c643e04cfa89b7cfb1f6f314f4ab22b92ade7b7b/modules/bowtie2.cfmod.pl#L33

This only works if you already have the software installed as environment modules, and those names match what CF expects. You can turn off this behaviour with the config options @ignore_modules (docs) and change the names with @environment_module_alias (docs).

If not using environment modules, then you need to have the software available on your system.

In our group we have migrated to using nextflow instead of Cluster Flow, which has built in support for conda, singularity and docker. As such, pipelines such as the ones that we're working on at https://nf-co.re/ do automatically install the required software for you.

I hope this helps!

Phil