This repository makes available the scripts used to download or build the GROND databases described in the manuscipt and available for download here.
For a more in-depth comparison of 16S-ITS-23S analysis methods, see our preprint (currently under review).
Please get in touch if you have any comments, issues, or suggestions for improvements.
I plan to update the database in line with each major GTDB release.
If I am behind the times and don't have a database for the latest GTDB release then let me know by posting an issue.
Database Download
Dependencies
Usage
Note on Database Construction
Note on Naming
The database is hosted on Zenodo so downloading each file individually is relatively simple.
The simplest method is to use the download button on the website.
For downloading with wget or curl - you can copy the URL of each file, minus the ?download=1 suffix.
For the sake of convenience, download scripts have been included in this repository - they just contain the curl commands ready to go and can be used by simply running sh download_gtdb_nr.sh.
Make sure these are in your $PATH
| Software | Version Tested |
|---|---|
| Barrnap | 0.9 |
| BBTools | 38.90 |
| BEDTools | 2.30.0 |
| csvtk | 0.24.0 |
| R | 4.1.0 |
| SeqKit | 2.2.0 |
| TaxonKit | 0.8.0 |
| VSEARCH | 2.17.1 |
| Package | Version Tested |
|---|---|
| Tidyverse | 1.3.1 |
| Ape | 5.0 |
The conda environment that I used to build the database can be created using conda env create -f grond.yml (the yml is provded in this repository).
This will create an environment called grond which can be loaded with conda activate grond.
This environment contains everything except csvtk and Seqkit. These are available through bioconda but I used preexisting non-conda installations.
refseq_grond.sh and gtdb_grond.sh are used to build databases using RefSeq and GTDB taxonomy respectively.
By default they will use:
These settings, along with VSEARCH clustering identity (default: 99.9%) and output directory (default: ./Grond_GTDB/ or ./Grond_RefSeq/) can be changed by editing the variables at the beginning of the relevant script.
Both scripts mentioned above will do the majority of the work but will call R scripts to identify rrn operons, perform quality checking (the details of which can be found in the GROND publication), and join rrn operon identifiers to the source genome taxonomy. R scripts are also used by the GTDB database builder to extract assembly information and identify genomes which are missing annotation.
Also included in the Scripts/ folder are scripts to:
GROND is envisaged as a tool to aid standardisation of 16S-ITS-23S rRNA analysis and allow comparison of results and, as such, building your own version would defeat the purpose.
That said, if you want to build your own version using the NCBI taxonomy system, make sure you have the most up-to-date version of the taxonomy database. I do this using the commands described in the TaxonKit manual.
This does not download automatically as I often encounter an EOF error while extracting the taxdump tarball so it is less hassle to do it manually using the accomanying script get_taxonkit_DB.sh before running refseq_grond.sh.
You do not need to do this when using the GTDB taxonomy system as the information is available on the website and will be automatically downloaded when running gtdb_grond.sh.
GROND is named after the wolf-shaped battering ram that broke the Great Gate of Minas Tirith during the Battle of the Pelennor Fields, itself named in homage to the great hammer of Angband wielded by Morgoth, the first Dark Lord.
It was originally named FANGORN (Full-length Amplicons for the Next Generation Of rRNa analysis) but was renamed to avoid confusion with the much more popular phangorn.
There may still be historic references to FANGORN in the scripts - please reach out if you find these and I will fix them