Installation

System requirements

RTMet requires:

  • A Unix-like operating system: macOS, Linux, BSD.

  • A distribution of the Conda package manager.

  • GNU coreutils.

  • git, curl, ssh and rsync

Some optional features will require:

  • An InfluxDB instance

  • JupyterHub

If you’re using MacOS, follow the instructions from Cylc documentation explaining additional dependencies.

Installing the workflow

Conda

If you don’t already have a distribution of Conda installed, we recommend installing Miniforge:

$ curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
$ bash Miniforge3-$(uname)-$(uname -m).sh

If it’s not already set, make sure that the Conda base environment does not activate by default:

$ conda config --set auto_activate_base false

Cloning the project

Download the workflow from the project’s repository on GitHub, to a location of your choice:

$ git clone https://github.com/MetaboHUB-MetaToul-FluxoMet/RTMet.git

We also have to make sure that Cylc finds the workflow by symlinking it:

$ mkdir ~/cylc-src
$ ln -s $(pwd)/RTMet/cylc-src/bioreactor-workflow ~/cylc-src/bioreactor-workflow

Setting up Cylc and its wrapper script

Cylc is a workflow manager and the core dependency of RTMet. Install it using the cylc.yml Conda environment file:

$ cd RTMet/cylc-src/bioreactor-workflow/envs
$ conda env create -f cylc.yml

Once it’s done, you’ll need to setup the Cylc wrapper script, which is needed for Cylc and Conda to work together.

We recommend installing it in a directory that is in your $PATH, such as /usr/local/bin (will require sudo access):

$ WRAPPER_DIR='/usr/local/bin'
$ conda activate cylc
$ sudo $(which cylc) get-resources cylc ${WRAPPER_DIR} && chmod +x ${WRAPPER_DIR}/cylc
$ sudo ln -s ${WRAPPER_DIR}/cylc ${WRAPPER_DIR}/rose
$ conda deactivate

Then, you’ll need to edit the wrapper script to point to the Conda environment where Cylc is installed.

- CYLC_HOME_ROOT="${CYLC_HOME_ROOT:-/opt}"
+ CYLC_HOME_ROOT="${CYLC_HOME_ROOT:-${HOME}/miniforge3/envs}"

To test your installation, launch the cylc command without any conda env active:

$ for i in $(seq ${CONDA_SHLVL}); do conda deactivate; done
$ cylc help

Installing workflow tasks environments

Bioinformatics tools are installed in separate Conda environments, for isolation and reproductibility purposes. For binneR, you’ll need to install it from the R console:

$ for file in wf-*.yml; do conda env create -f $file; done
$ conda activate wf-binner && Rscript -e "remotes::install_github('aberHRML/binneR', upgrade='never')"
$ conda deactivate

Optional: Installing InfluxDB

Visualizing results in real time requires an InfluxDB instance. One option is to use InfluxData’s official cloud solution . It’s quick and easy to set up, so we recommend it for testing the workflow.

Warning

The free-tier of InfluxData’s cloud solution is quite limited. Buckets have a retention policy of 30 days, which means your data will be deleted one month after it being uploaded. Don’t use it as a primary backup.

The other one is to use the self-hosted version, InfluxDB OSS v2. See installation instructions from InfluxDB’s documentation . Either way, make sure to setup your InfluxDB instance by creating an organization and a first user.