Instructions for class exercise

05-workflows/snakemake/README.md

@@ -1,16 +1,43 @@
-## Hands-on Exercise: Running Snakemake workflow
+# Hands-on Exercise: Running Snakemake workflow
 
 This is a hands-on exercise on running Snakemake workflows on the cluster (either SCG or Sherlock). Through a dummy workflow, students will learn the basics of a Snakemake workflow: rules; wildcards; resource management; environment management. 
 
-### Snakemake cli commands
+We will run `snakefile01.smk` and `snakefile03.smk` during class. (We will just skim `snakefile02.smk`.
+
+## Prerequisites
+* Installed snakemake v7.32.2 on scg/sherlock
+  * If you don't have conda on the cluster, install miniconda. Follow the instructions under [here]([https://github.com/bhattlab/scg_tools](https://github.com/bhattlab/scg_tools?tab=readme-ov-file#setting-up-your-environment-with-conda)). Make sure you've downloaded conda by trying `which python` and `which conda`.
+  * Follow the instructions in this [link](https://snakemake.readthedocs.io/en/v7.32.2/getting_started/installation.html), making sure to specify the version [hint: use `snakemake=7.32.2`]
+* Clone this repo to a dir on the cluster [hint: use `git clone`]
+
+## Class exercise
+You can choose either to run locally, or to send jobs (i.e. sbatch). All of this should be handled by snakemake profiles. You should try both!
+
 To run locally, just provide no. of cores e.g.
 `snakemake --snakefile <snakefile.smk> -c1`
 
 To use slurm, use one of the snakemake profiles, depending if you're using scg/sherlock (feel free to change e.g. partition) e.g.
 `snakemake --snakefile <snakefile.smk> --profile <snakemake_profiles/slurm_sherlock>`
 
+If you use `snakemake_profiles`, please inspect the `config.yaml` file and `chmod u+x status-sacct.sh` 
+* Make sure the paths in `config.yaml` makes sense and you understand what it's there to do
+* `status-sacct.sh` is just a script to ask slurm the status of each job. For our class, this is a bit over the top. This makes sure we're aware of jobs that silently fails (the exact `sacct` status we're monitoring are in the script)
+
 To do a dry-run (you should!), just put `-n` at the end, e.g. 
 `snakemake --snakefile <snakefile.smk> -c1 -n`
 
-If you use `snakemake_profiles`, please inspect the `config.yaml` file and `chmod u+x status-sacct.sh` 
-* This is just a script to ask slurm the status of each job. For our class, this is a bit over the top. This makes sure we're aware of jobs that silently fails (the exact `sacct` status we're monitoring are in the script)
+
+Please follow the following instructions:
+
+### `snakefile01.smk`
+Try:
+1. Review and check it makes sense. Run this as-is on scg/sherlock 
+2. Edit `snakefile01.smk` to stop at `rule trim`
+3. Edit this workflow to change default resources
+
+### `snakefile03.smk`
+1. Review and check it makes sense. Run this as-is on scg/sherlock
+2. Try deleting intermediate files. What happens when you rerun the workflow? 
+
+
+