We have set some sensible defaults for Hecatomb on the following minimum recommended system requirements:
- 32 CPUs (or 16 CPUs with hyperthreading)
- 64 GB of RAM
- approximately 55 GB HDD space (for the databases)
- Additional HDD space for temporary and output files (highly dependent on input file sizes)
Larger datasets may require more RAM (and CPUs to speed things along). As such it is highly recommended to run Hecatomb on a HPC cluster.
Hecatomb was developed as a set of Snakemake workflows, which are controlled by a python launcher for your convenience. While the pipeline utilises a number of other programs and tools to run, these are all managed by the pipeline using conda and mamba.
Install Hecatomb via conda:
# This command will create a new conda env called 'hecatomb' and will install Hecatomb and all of it's dependencies. conda create -n hecatomb -c conda-forge -c bioconda hecatomb
# To use Hecatomb, activate your new conda env. conda activate hecatomb # Check that it's installed hecatomb -h
If you're running Hecatomb on a cluster it is highly recommended to use Snakemake profiles (Set up a profile for Snakemake).
You may also want to customise the available resources for your system, whether you're using a cluster or running locally (Advanced configuration).
Download the databases
Before running Hecatomb for the first time you will need to download the databases. You will only need to do this step once (unless we update the databases). You can rerun this step as much as you like; the pipeline will only download any database files that are missing.
# Either, run locally hecatomb install # Or, for a HPC cluster using a Snakemake profile (replace 'slurm' with your profile name) hecatomb install --profile slurm
Run the test dataset
Hecatomb comes with a test dataset that you can run which will take a few hours to complete.
--test in place of specifying your read directory with
# run locally hecatomb run --test # run on cluster using a Snakemake profile hecatomb run --test --profile slurm
Build the docs
These document pages can be built from the repo like so.
# install mkdocs pip install mkdocs # cd to your install directory cd /path/to/hecatomb # build mkdocs build