buttery-eel v0.8.2

What's Changed

• pi bug in resume by @Psy-Fer in #80
• fix reader file location bug by @Psy-Fer in #81

Full Changelog: v0.8.1...v0.8.2

buttery-eel v0.8.1

Some fixes to timeout on initial start up for slow HPC systems

Fix GPU detection for multiple GPUs or selecting something other than cuda:0/cuda:all

Full Changelog: v0.8.0...v0.8.1

buttery-eel v0.8.0

Pre-release: has errors, use v0.8.1 or higher

This release bring compatibility with dorado-server-7.11.2

The last few dorado-server releases had a lot of changes to the interface, and so are not stable to use with buttery-eel.
The most major change for users is the removal of --config and its replacement with --model and --modbase_models

The GPU is now printed in the fastq output
The way data is fed to the GPU has been revamped to maximise GPU usage due to internal changes with how dorado-server works.
A few other tweaks that can be seen in the full changelog.

What's Changed

• New data model and compatibility with dorado-server-7.11.2 by @Psy-Fer in #79

Full Changelog: v0.7.2...v0.8.0

buttery-eel v0.7.2

This is mostly a bug fix patch.

There were some strange timeout errors occuring with ultra long kits where batches with megabase reads were causing the ram of a machine to spike, leading to an OOM error. If you are working with ultra long reads, please reduce the --slow5_batchsize to 2000 and keep the --procs to 10 or less. This should keep ram usage down.

What's Changed

• v0.7.2 fixing timout bugs and resume file handling by @Psy-Fer in #74

• move timeout check to write worker
• increase timeout to 5000 seconds
• fix samples/s bug by changing total sample value to u64
• handle files on resume, appending to seq_sums and error on opening a fastq/sam that already exists

Full Changelog: v0.7.1...v0.7.2

buttery-eel v0.7.1

This patch fixes an issue when running buttery-eel with dorado-server versions older than 7.6.8

• fix namespace error on estimate_poly_a by doing a version check

Full Changelog: v0.7.0...v0.7.1

buttery-eel v0.7.0

This version has a bug when running with dorado-server versions lower than 7.6.8 - Please use v0.7.1 or newer

buttery-eel v0.7.0 significantly updates the error model.
With the new --resume option, runs that did not finish can be continued without re-basecalling reads already processed.

This release has been tested up to dorado-server-7.6.8

What's Changed

• better error handling
• --resume to resume a crashed run by providing the fastq or sam file of the crashed run
• --dorado_model_path to work like dorado model paths
• --estimate_poly_a polyA/T tail estimation
• samples per second reported at the end of a run
• qscores reported as floats instead of ints and samtag changed to qs:f: (alignment with latest dorado changes)
• update split script to handle floats and legacy ints
• updated docs

Pull request:

• Handle errors better and add new features by @Psy-Fer in #72

Full Changelog: v0.6.0...v0.7.0

buttery-eel v0.6.0

What's Changed

• RNA fixes and u2t conversion by @Psy-Fer in #61
• thread model doc update
• test updates and fixes for new test machine

Full Changelog: v0.5.1...v0.6.0

buttery-eel v0.5.1

What's Changed

• add dot try on barcode summary by @Psy-Fer in #49

Full Changelog: v0.5.0...v0.5.1

buttery-eel v0.5.0

What's Changed

• 7.4.12 dorado for v0.5.0 buttery-eel by @Psy-Fer in #47
• Added more SAM tags
• Breaking API changes accounted for - see README
• Prints total reads basecalled at the end
• Skip queue introduced and will write a skip file if skip reads are detected
• macOS changes to the queue library so it works
• read_splitting on by default for ont lib >=7.4.12

Full Changelog: v0.4.3...v0.5.0

Buttery-eel v0.4.3

What's Changed

Fixed demux parent/read ID issue
major code restructure and compatibility with lastest dorado server lib name change: ont-pybasecall-client-lib
prep for Duplex calling, but currently unavailable due to bugs/missing methods in ONT library.

• Lib change and prep for duplex by @Psy-Fer in #39

Full Changelog: v0.4.2...v0.4.3