v0.7.4

• updated the score-based validation workflow to apply SVM abstention on all remaining reads

v0.7.3

• Add a CLI command to run the full XspecT pipeline
• Add workflows for pan genome-informed training, score-based validation using SVM
• Update benchmark & benchmark docs

v0.7.2

Added an option that allows users to exclude species by ID during classification

v0.7.1

• Lazy load PubMLST available organisms to avoid rate limiting issues
• Fix docs CI error

v0.7.0

• Refactor MLST classification to better align with XspecT model architecture
• Update benchmark, e. g. to use Refseq instead of Genbank, use a fixed version of XspecT, simulate reads using ART and use all available genomes for read generation
• Update docs
• Minor bugfixes

v0.6.0

• Update benchmark
• Add misclassification detection
• Expose data quality filters to the CLI when training with NCBI data
• Update docs

v0.5.4

• Updated the benchmark to use 100000 reads of length 100 for each input assembly, with up to three assemblies per species
• Speed up CLI by lazy loading dependencies
• Optionally show display names in classification results using -n / --display-names
• Updated docs

v0.5.3

Bugfix: Ensure that the subtree is also considered when getting accessions

v0.5.2

• Filtering capabilities are now available in XspecT Web
• Output parent directories are now automatically created, if they do not exist
• Documentation updates (Mkdocs & docstrings)
• Minor changes & bugfixes

v0.5.1

• Support for whole directories as input
• Allow saving of classification results when filtering
• More comprehensive logging
• Input validation for filtering thresholds
• Search for XspecT data directory in home directory, not just in the current working directory
• Updated readme