v0.11.0

Feature release

• Consensus models for guppy 3.3 and 3.4.
• Aggregation of Guppy modified base probability tables.
• Multi-thread stitching of inference chunks in medaka_consensus.
• Optionally run whatshap phase at the end of medaka_variant.

Please do not attempt to use the "Source code" assets attached below for running medaka, they do not include medaka's model files.

v0.10.1

Minor fix release

• Fix bug where feature matrix was misaligned with coordinate system.
• Add missing arguments from smolecule command.
• Output contig names are no longer written as samtools-style regions.
• Fixed issue with medaka_variant failing on zero-coverage regions.
• Rename incorrectly named diploid SNP calling model.
• Made variant calling faster by resolving trivial bottleneck in variant classification.

Please do not attempt to use the "Source code" assets attached below for running medaka, they do not include medaka's model files.

v0.10.0

Feature release

• Switched variant calling to an explicitly diploid calling model.
• Added a -f force overwrite option to medaka_consenses.
• Refreshed E. coli benchmark to include effect of racon.
• Refreshed variant calling benchmarks.
• Added C. elegans assembly benchmarks to documentation.
• Fixed bug causing larger than requested overlap in inference chunks.
• Corrected parsing of region strings with multiple : charaters
• Fixed rare consensus stitching error.

Please do not attempt to use the "Source code" assets attached below for running medaka, they do not include medaka's model files.

v0.9.2

Minor fix release.

• Additional fix to handling lowercase reference sequences.
• Fix bug in creation of RLE alignments.
• Update update_model.py script.
• Remove option to select labelling scheme during training.
• Unify how LabelSchemes store training data.

Please do not attempt to use the "Source code" assets attached below for running medaka, they do not include medaka's model files.

v0.9.1

Minor fix release

• Fix regression in medaka stitch and medaka snp speed.
• Handle lowercase letters in reference sequences.
• Remove dill and yaml requirements.

Please do not attempt to use the "Source code" assets attached below for running medaka, they do not include medaka's model files.

v0.9.0

Bugfix and training refactor release

• Fix readlink issue on MacOS
• Fix bug where medaka_variant did not call indels by default
• Fix bug in determining when to split contigs
• Drop support for older basecaller models (guppy<3.0.3)
• Store models in git-lfs
• Simplify medaka_variant workflow for speed
• Make network feature generation 2x faster
• Add smolecule command
• Log use of GPU and cuDNN, noting workaround for RTX cards
• Refactor labelling of training data and storing of models
• Reimplement RLE feature generation

Please do not attempt to use the "Source code" assets attached below for running medaka, they do not include medaka's model files.

v0.8.2

Documentation release

• Clarify suggested workflows in documentation.

v0.8.1

Patch release

• Patch import of loading of older models

v0.8.0

v0.8.0

Model release

• Add support for R10 basecaller
• Add support for diploid multi-labelling models
• Upgrade to tensorflow 1.14.0

Note: pypi binaries for the updated tensorflow are linked to CUDA10, previous medaka releases used a tensorflow version that was linked to CUDA9. This only affects users who run medaka on GPU hardware.

v0.7.1

Bug fix release

• Fix regression in consensus stitching when chunks do not overlap