add information about the way of writing workflow

workflows/README.md

+# Snakemake workflows
+
+This directory contains every workflows built from Snakemake tool found in the `tools` directory of the repository.
+
+## Design and best practices
+
+All the Snakemake workflows try to respect some rules and best practices in their design.
+Therefore, workflows have the following parts:
+
+1. reference to `config.yaml` file
+2. reference to tool rules
+3. set up of common variable for the workflow
+4. set up of specific variables for the tools
+5. `rule all` to specify what is/are the specific file(s) expected from the workflow.
+
+#### Reference to tool rules
+
+The location of the tools included in the workflow can vary depending your organization.
+Therefore, we decide to put every references at the begining of the workflow to make it easy
+to customize and change if necessary.
+
+#### Set up common variable for the workflow
+
+Then every common variables are set up. This can be handy for reference to a common directory.
+
+#### Set up specific variables for the tools
+
+Every specific variables for every tools are then specified. For more details about the way
+every tool is describe, you can refer to the `README` of the `tools` directory of the
+repository.
+
+In brief, every tool has its input and output refered as a variable: `__TOOLNAME_variable`.
+Therefore every variable needs to be configured and we recommand using reference to a `config.yaml` file.
+Thus, we can easily set up the input of a tool `B` as being the output of `A`:
+
+```python
+# ---- Tool A
+__toola_input = config.get('input')
+__toola_output = "file.a"
+
+# ---- Tool B
+__toolb_input = __toola_output
+__toolb_ouput = config.get('output')
+```
+
+## Usage
+
+### Using on TARS
+
+For the following examples, we consider that we want to run every job a dedicated queue called `atm`.
+
+First need a virtual environment with snakemake:
+
+```bash
+module load Python/3.6.0  # Install seems to fail with latest version of Python
+python3 -m venv .snakemake_venv
+source .snakemake_venv/bin/activate
+(.snakemake_venv)$ pip install snakemake
+```
+
+Then, to run a `SnakeFile` on Tars we use a command similar to the following:
+
+```bash
+sbatch --qos=atm -p atm -c 1 snakemake -p -j 6 --cluster-config cluster.yml --cluster "sbatch --qos={cluster.queue} -p {cluster.queue} -c {threads}"
+```
\ No newline at end of file

workflows/cluster.yml

+__default__:
+  queue: atm
\ No newline at end of file