DNAKit 0.1.4

DNAKit 0.1.4 adds reference-based gene annotation, general-purpose LLM calls, custom DNA property training, traditional machine-learning models, and classification/regression distribution plots.

• Add Python bioinformatics APIs for encodings, metrics, phylogenetics, statistics, cloning, optimization, workflows, and optional external backends.
• Add minimap2-based gene annotation with reference preparation and Python/CLI interfaces.
• Add synchronous/asynchronous LLM calls, provider configuration, and bounded file inputs.
• Add custom property training with linear probing, supported full fine-tuning, and traditional machine-learning estimators.
• Add 27 fine-tuned checkpoint classifiers and raw Levenshtein diversity/novelty metrics since 0.1.3.
• Expand bilingual documentation and examples, move development resources under others/, and update release packaging checks.

Validation: 1,525 local tests passed, 38 optional tests skipped; both tutorial notebooks and all 15 correctness checks passed. Ruff formatting/lint, strict documentation build, distribution metadata, and archive audits passed. GitHub CI passed on Python 3.10–3.14, and the TestPyPI release gate passed. Reviewed TestPyPI hashes and isolated installation were verified. Mypy remains advisory and reports 62 issues in the local environment.

Install:

python -m pip install --upgrade dnakit==0.1.4