Running the workflow on Uppmax¶

If you will be running the workflow on the Uppmax HPC clusters here are some helpful tips.

Setting up database files¶

All resources required to run this workflow are automatically downloaded as needed when the workflow is executed. However, some of these (often large) files are already installed in a central location on the system at /sw/data. This means that you can make use of them for your workflow runs thereby saving you time and reducing overall disk-usage for the system.

First create a resource/ directory inside the directory where you will be running the workflow:

mkdir resources

eggNOG database¶

The 5.0 version of the eggNOG database is installed in a central location on Uppmax under /sw/data/uppnex/eggNOG/5.0. To make use of this database simply symlink the files in the directory into a directory resources/eggnog-mapper:

mkdir resources/eggnog-mapper
ln -s /sw/data/eggNOG/5.0/* resources/eggnog-mapper
head -1 /sw/data/eggNOG/eggNOG_data-5.0_install-README.md > resources/eggnog-mapper/eggnog.version
touch resources/eggnog-mapper/download.log

Pfam database¶

To use the Pfam database from the central location, create a pfam sub-directory under resources and link the necessary files from the central location, run the following:

mkdir resources/pfam
ln -s /sw/data/Pfam/31.0/Pfam-A.hmm* resources/pfam/
cat /sw/data/Pfam/31.0/Pfam.version > resources/pfam/Pfam-A.version

This installs the necessary files for release 31.0. Check the directories under /sw/data/Pfam/ to see available release versions.

Kraken database¶

For kraken there are a number of databases installed under /sw/data/Kraken2. Snapshots of the standard, nt, rdp, silva and greengenes indices are installed on a monthly basis. To use the latest version of the standard index, do the following:

Create a sub-directory and link the index files

mkdir -p resources/kraken/standard
ln -s /sw/data/Kraken2/latest/*.k2d resources/kraken/standard/

2. From a reproducibility perspective it’s essential to keep track of when the index was created, so generate a version file inside your kraken directory by running:

file /sw/data/Kraken2/latest | egrep -o "[0-9]{8}\-[0-9]{6}" > resources/kraken/standard/kraken.version

Modify your config file so that it contains:

classification:
    kraken: True

kraken:
    # generate the standard kraken database?
    standard_db: True

Non-standard databases¶

Using any of the other non-standard databases from the central location is also a simple process, e.g. for the latest SILVA index:

mkdir -p resources/kraken/silva
ln -s /sw/data/Kraken2/latest_silva/*.k2d resources/kraken/silva/
file /sw/data/Kraken2/latest_silva | egrep -o "[0-9]{8}\-[0-9]{6}" > resources/kraken/silva/kraken.version

Then update your config file with:

kraken:
  standard_db: False
  custom: "resources/kraken/silva"

Centrifuge database¶

There are several centrifuge indices on Uppmax located at /sw/data/uppnex/Centrifuge-indices/20180720/, but keep in mind that they are from 2018.

The available indices are: p_compressed, p_compressed+h+v, p+h+v, and p+v (see the centrifuge manual for information on what these contain).

To use the p_compressed index, run:

mkdir -p resources/centrifuge/p_compressed/
ln -s /sw/data/Centrifuge-indices/20180720/p_compressed.*.cf resources/centrifuge/p_compressed/

Then update your config file to contain:

classification:
    centrifuge: True

centrifuge:
    custom: "resources/centrifuge/p_compressed/p_compressed"

GTDB¶

To use the centrally installed Genome Taxonomy Database (GTDB) release on Uppmax, do:

mkdir -p resources/gtdb
ln -s /sw/data/GTDB/R04-RS89/rackham/release89/* resources/gtdb/

Then make sure your config file contains:

binning:
    gtdbtk: True

nr¶

Uppmax provides monthly snapshots of the nr non-redundant database. While the formatted file cannot be used directly with the nbis-meta workflow you can save time by making use of the already downloaded fasta file.

To use the latest snapshot of nr for taxonomic annotation of contigs, do:

mkdir resources/nr
ln -s /sw/data/diamond_databases/Blast/latest/download/nr.gz resources/nr/nr.fasta.gz
file /sw/data/diamond_databases/Blast/latest | egrep -o "[0-9]{8}-[0-9]{6}" > resources/nr/nr.version

Then update the taxonomy section in your config file to use the nr database:

taxonomy:
    database: "nr"

Uniref90¶

The UniRef90 database is clustered at 90% sequence identity and Uppmax provides downloaded fasta files that can be used directly with the workflow.

To use the latest snapshot of UniRef90 for taxonomic annotation of contigs, do:

mkdir resources/uniref90
ln -s /sw/data/diamond_databases/UniRef90/latest/download/uniref90.fasta.gz resources/uniref90/uniref90.fasta.gz
file /sw/data/diamond_databases/UniRef90/latest | egrep -o "[0-9]{4}_[0-9]{2}" > resources/uniref90/uniref90.version

Then update the taxonomy section in your config file to use the uniref90 database:

taxonomy:
    database: "uniref90"

Configure workflow for the SLURM Workload Manager¶

The workflow comes with the SLURM snakemake profile pre-installed. All you have to do is to modify the config/cluster.yaml file and insert your cluster account ID:

__default__:
    account: staff # <-- exchange staff with your SLURM account id

Then you can run the workflow with --profile slurm from the root of the workflow directory, e.g.:

snakemake --use-conda --profile slurm -j 100 --configfile myconfig.yaml

Here the -j 100 flag means that snakemake can have at most 100 jobs in the queue at the same time for this run.