Oliver Schwengers

Our 9th @denbi.bsky.social "Microbial Genomics training course" is now open for registration!

3 sessions full of microbial bioinformatics at JLU Giessen:

I: QC & QA, assembly
II: regional & functional annotation
III: comparative genomics

Info & registration: www.denbi.de/training-cou...

100% agreed - exactly that was also our line of thinking.

Thanks @kblin.bsky.social for pointing out. Of course, it is not a proper encryption, however hard to guess.

So, please use a proper encryption for sharing - if required.

Even without registration, results can easily and safely be shared with others using encrypted URLs.

Of course, your data is handled with care automatically discarded after 30 days.

Just give it a try: bakta.computational.bio
(5/5)

Results are presented via a cross-referenced data table, a linear genome browser and circular genome plots - just choose your viz!

Of course, all results are also available for download in various standard file formats: GFF3, GenBank, EMBL, FASTA, TSV, TXT, JSON…
(4/5)

All executions are conducted in a scalable Kubernetes backend hosted within the @denbiOffice cloud.

By this, Bakta Web dynamically scales out to meet temporal peak demands.
(3/5)

Bakta Web is our web application accompanying Bakta’s CLI version.

It provides interactive widgets supporting users with both, the provision of input data and sample/sequence metadata, and the interpretation of annotation results.
(2/5)

Bakta Web – rapid and standardized genome annotation on scalable infrastructures Abstract. The Bakta command line application is widely used and one of the most established tools for bacterial genome annotation. It balances comprehensiv

We happily present: “Bakta Web – rapid and standardized genome annotation on scalable infrastructures” @OxUniPress NAR’s Web Server issue
doi.org/10.1093/nar/...

Easy to use, no registration, fast, scalable, various visualizations, in sync with Bakta CLI:
bakta.computational.bio
(1/5)

Interested in bacterial genomes?

Hundreds of thousands, even millions?

All annotated, taxonomically classified, integrated with metadata.

Easily searchable, viewable, downloadable, in sync with #AllTheBacteria.

Then BakRep is for you! Poster P-CM-102 @vaam-microbes.bsky.social #VAAM25

Of course, the DB light version was also updated, however DB sizes have barely changed.

(6/6)

Release v1.11 - Database ups and downs · oschwengers/bakta This is the eleventh minor release (v1.11) introducing database schema version 6. Compatible database scheme version: 6 Important From this release on, Bakta will use xz instead of gz for all data...

This introduced a DB major scheme update from v5 to v6, which in turned resulted in a new software release v1.11.0:

github.com/oschwengers/...

(5/6)

Due to this massive increase in known protein sequences, DB size has grown from 72GB to 83GB (deflated).

To mitigate negative download effects for our users, we therefore switched tarball compressions from gzip to xz.

Thus, compressed DB size actually decreased from 37GB to 30GB!

(4/6)

To further improve annotation quality, external annotation sources have also been updated:

- COG: 2020 -> 2024
- Pfam: 36 -> 37.2
- RefSeq: r220 -> r228
- Rfam: 14.10 -> 15
- UniProtKB: 2023_05 -> 2025_01

Also, NCBI AMRFinderPlus and VFDB have been updated to the most recent versions.

(3/6)

This is a massive increase in known protein sequences - which is great!

Numbers of IPS and PSC have grown by 22% and 13%.

(2/6)

🦠🧬🖥️ New Bakta DB v6.0 released!

After a year, it was time for a Bakta database update - and it's a huge one:
- IPS: 330.9M
- PSC: 135.3M
- PSCC: 37M

doi.org/10.5281/zeno...

👇 1/6

Our paper on bacterial #Resistance prediction with #Genomics and #MachineLearning reached a milestone with its 120th citation! 🎉 @oschwengers.bsky.social

academic.oup.com/bioinformati...

Last week before Christmas...
Though not being a captain, that's the post, ... just that ;-)

I just made a first release of TNA - new tool to compare 2 bacterial genomes. Is shamelessly inspired by ACT, but should be simpler to use. Drag-n-drop two genomes: it runs BLAST for you and then you can visualize. For Mac, Windows 11, Linux tna.readthedocs.io

PubMLST - Public databases for molecular typing and microbial genome diversity Public databases for molecular typing and microbial genome diversity

BIGSdb v1.49.1 has been released. This version improves the layout of user registration and admin pages, and provides the option to export LIN code prefixes of specified length in the Export plugin. BIGSdb powers the databases on pubmlst.org and bigsdb.pasteur.fr.

github.com/kjolley/BIGS...

Info: We updated our Bakta WEB version:

- MUCH faster annotations via optimized Kubernetes nodes and locally-attached databases
- Software-Update to v1.10.1
- Database-Update to v5.1

Publicly available at: bakta.computational.bio

Release v1.10.1 · oschwengers/bakta This is the first v1.10 patch release (v1.10.1). Improvements Improved handling of draft genome in circular plots: 0118983 Added N90 to genome statistics: 7abfa9a Fixes Fixed changed Micromamba ...

Community contributions:
Last but not least, I'd like to thank all users & contributors for incredible feedback, ideas and comments.

Just a few hours after v1.10, reports, fixes and contributions helped to patch -> v1.10.1 - absolutely amazing!

github.com/oschwengers/...

Of note:
We updated Bakta using AMRFinderPlus v4.0.3 that introduced a DB change. So, Bakta might crash using existing DBs.

To update/fix this, please, execute the update command Bakta prints to the CLI - or just re-download it:
'bakta_db download'.
(10/11)

A hard decision, but to streamline further maintenance and allow for novel features, we have to go on…
So, since v1.10 Bakta cannot annotate signal peptides anymore. If you need this feature, please make sure to use Bakta v1.9.4.
(9/11)

Python dependency and DeepSig support:
Unfortunately, we had to bump Python dependency to >=3.9 thus, lost support of DeepSig, which is pinned to Python 3.8 and is not supported anymore.
(8/11)

Further improvements:
- translation table 25 support
- export of CRISPR/tRNA seqs
- improved INSDC compliance
- new parameter '--locus-tag-increment'
- revised designation of truncated features and (real) pseudogenes
- 3rd party dependency updates

… and important bug fixes.
(7/11)

Genome plots:
To improve Bakta’s circular genome plots, we replaced good old Circos by the novel pure-Python pyCirclize library. Thus, we could also add new features, such as legends and new 'bakta_plot' parameters:
bakta_plot --type cog –dpi 300 --size 8 --label='my custom label text'
(6/11)

Overlap filters:
Sometimes, certain features (mostly CDS/sORFs) are filtered-out as they overlap with others. If these are of interest, now these can be kept by skipping all overlap filters via the new '--skip-filter' option.
(5/11)

As requested from several users, Bakta now writes all annotation metrics (coverage, sequence identities, bitscores, etc) to a dedicated TSV file for further inspection. (4/11)

Recovery of result files:
Bakta’s output too large? Via a new 'bakta_io' cmd all result files can be recovered from its json[.gz] file within seconds.
By this, ~84 Mb of result files (e.g. E.coli) are stored in a 5.8 Mb .json.gz file reducing storage requirements by 93%.
(3/11)

User-provided HMMs:
Just like '--proteins' for user-provided protein sequences, Bakta now accepts trusted HMMs via a novel '--hmms' parameter and uses them in its internal expert annotation module.
(2/11)