Why?

For those thinking "Why did you make this repo?", this is a collection of 'short' snakemake workflows that may fulfill a single portion of a larger workflow. The idea is to have a set of easily run workflows that build off each others output.

For you thinking "Still, why?", this will allow for multiple positive outcomes with only minor inconvenience.

1. Smaller workflows equal smaller DAGs.

• For those of us who have hit 20,000+ DAGs regularly, the workflow do not always run and can take a significant time to begin. The shorter workflow of only a few step bypasses this limitation. Simply run one workflow after the next until the grand workflow is complete.

2. Focusing the workflows on individual tasks (i.e. quickly make sourmash signature or a spacegraphcats catlas), I aim to fight bloated workflows as much as possible.

• Ideally, each workflow will fall into one of four categories:
  1. Acquire raw data
  2. Process to refine data
  3. Acquire necessary results
  4. Report results in standard format

3. Smaller workflows are easier to debug.

• Perhaps this is a personal issue, but I like debugging the shorter workflows. The longer workflows break at rule 1 then rule 5 then at rule ... You get the idea. Here the workflows are individualized, this allows the user to narrow in on the broken part immediately. Debugging will also be easier with a modular approach. As long as all the input files exist for the first rule in each modular workflow, the entire grand workflow may be run to completion with only the buggy workflows breaking.

4. Smaller workflows may be built upon as a seed template.

• Well, the idea here is that you like longer workflows and want to take the workflows from this repo as a beginning. A starting point on your adventure into workflow writing. Cool, do your do, you do you.

dotfiles

Configuration files for running snakemake with the --profile argument. This allows dynamic workflows to utilizing HPC resources in a holistic way.

config.yaml

This is the configuration of the --profile. It contains default allocations for both sbatch and snakemake.

Note: This dotfile configuration breaks with snakemake version 8. Use snakemake version 7.32.4 until further notice.

slurm-status.py

A script called by config.yaml for better cluster performance. See https://snakemake.readthedocs.io/en/stable/tutorial/additional_features.html#using-cluster-status for details.

workflows

Snakemake files for creating retrieving, sketching, gathering, and parsing data commonly used in the Data Intensive Biology lab at the University of California, Davis.

There is a directory housing the configuration and environment files called by the workflow. In addition, the metadata directory contains a SraRunTable.txt file for an example dataset from CAMI.

Sourmash workflows

Quickstart:

To begin, for ssh:

git clone git@github.com:dib-lab/portable-snakemake-workflows.git

For https:

git clone https://github.com/dib-lab/portable-snakemake-workflows.git

Consider using the command below immediately after git clone to correctly place the dotfiles for --profile arugments on an HPC using slurm:

( cd portable-snakemake-workflows && mkdir -p $HOME/.config/snakemake/slurm && mv dotfiles/config.yaml $HOME/.config/snakemake/slurm && mv dotfiles/slurm-status.py $HOME/.config/snakemake/slurm && cd - )

The following portable workflows will run sequentially.

1a. download-sketch

A set of fast, robust approaches to temporarily downloading sequence files from archives and sketch sourmash signatures.

Quickstart:

If the dotfiles have been properly installed, run:

snakemake -s download-sketch-cluster --resources allowed_jobs=100 --profile slurm

This will download and sketch the CAMI dataset into the sigs directory at scale of 1000 and kmer size of 21, 31, and 51. These values are set in the download-sketch-config.yaml and may be changed there at any time.

2a. extract-prefetch

An efficient approach to temporarily extract the sketches in sourmash signatures that are passed into the 'preferred' prefetch method. Creating a picklist for rapid sourmash gather results.

Quickstart:

(While on the Farm HPC) From within the scripts directory, run ./fastgather-db-submit-jobs.sh to create the proper structureed GTDB databases for fastgather to run.

If the dotfiles have been properly installed and download-sketch has been successfully ran, run:

snakemake -s extract-prefetch-cluster --resources allowed_jobs=100 --profile slurm

This will temporarily extract all k-sizes and create a prefetch csv file of the CAMI sketches in the gather/prefetch directory for the scale and k value set in the extract-prefetch-config.yaml.

1b. sbt-lca-databases

[WIP] A workflow to take any existing sourmash database and create a Sequence Bloom Tree (SBT) index and Lowest Common Ancestor (LCA) database.

This was curated using existing scripts from the repo -> https://github.com/dib-lab/2018-ncbi-lineages

An additional script, find-rank-kmers.py was included to extract the rank-specific kmers from an LCA database.

usage: find-rank-kmers.py [-h] -o OUTPUT [-s SAVE_NAMES] lca_db keep_rank

positional arguments:
  lca_db
  keep_rank             Keep the rank specified (i.e. Species)

optional arguments:
  -h, --help            show this help message and exit
  -o OUTPUT, --output OUTPUT
                        output signature containing rank-specific hash values
  -s SAVE_NAMES, --save-names SAVE_NAMES
                        save list of matching signature names from that lineage to this file

TODO:

1. create example dataset using sourmash sketch fromfile to run the workflow

• using a set of fasta genomes currently from refseq plastid

2. combine into single cmd with ( && )

• mkdir dbs
• cd dbs
• wget -c ftp://ftp.ncbi.nih.gov/refseq/release/plastid/plastid.3.1.genomic.fna.gz -o plastid.3.1.genomic.fna.gz
• sourmash sketch dna -p scaled=100,k=21,k=31,k=51 -p scaled=1000,k=21,k=31,k=51 --singleton plastid.3.1.genomic.fna.gz -o plastid.3.1.genomic.sig.gz
• ...
• cd ..
• snakemake -s sbt-lca-databases -j?