Installation¶

Install from PyPI¶

Install 0.1.5:

pip install dnakit==0.1.5

Package page: pypi.org/project/dnakit

Other platforms¶

The project repository has provided the following platform adaptation files:

These adaptation files are pinned to 0.1.1, and the official installation command needs to be reviewed and released by the corresponding platform before it can be used.

Development environment¶

Create and activate the complete Conda development environment in the project root directory:

conda env create -f environment-dev.yml
conda activate dnakit-dev

The development environment contains testing, type checking, documentation, and build tools. For additional dependencies on optional functions see pyproject.toml.

Optional bioinformatics backend¶

Install the optional backend used by the VEP, ClinVar, dbSNP, gnomAD, dN/dS, and Golden Gate functions:

python -m pip install "dnakit[external-tools]"

The default pip install dnakit neither installs nor loads this backend.

Deep-learning property prediction¶

Install the Enformer regulatory-track and NT Revised/Genomic Benchmarks task-checkpoint inference dependencies:

python -m pip install "dnakit[neural,neural-enformer]"

The 27 task checkpoints are not distributed in the package. Prediction automatically downloads the selected task checkpoint if absent and reuses existing files. The default cache is ./ckpt/enformer-benchmarks/; override it with checkpoint_dir or checkpoint_path. See prediction dependencies and checkpoint caching. Task examples are organized into regulatory tracks, epigenetic classification, and promoter and enhancer classification.

Agents and MCP¶

Install the MCP server used by Codex, Claude, Cursor, and other MCP-capable Agents:

python -m pip install "dnakit[agent]"

Run dnakit-mcp after installation. See LLM tools → MCP for configuration.

Repository¶

github.com/mapengsen/DNAKit

Optional dependencies for bioinformatics Python APIs¶

Install the current source checkout's common genomics capabilities with Python ≥3.10:

python -m pip install -e ".[genomics]"

The genomics extra installs NumPy, Pyrodigal, pysam and PyFastANI for genotype PCA, genomic relationships and breeding prediction, prokaryotic CDS prediction, BAM statistics and genome ANI. These APIs do not need separate PLINK, Prodigal, samtools or FastANI executables. See the genomics tutorial for their scope and examples.

These additions are not yet published to PyPI; after publication, use python -m pip install "dnakit[genomics]". Wheel availability depends on the dependency and platform; platforms without a matching wheel may need a compiler. This extra does not install all 43 external analyses, reference databases or model weights.

In the current workspace, use python -m pip install -e ".[bio]" for general bioinformatics Python dependencies, or install only .[bio-analysis]. Use .[genomics] above for the new genomics computing dependencies. External command adapters for MAFFT, BLAST, HMMER, BWA, samtools and bcftools still require their executables. See Dependencies and implementation.

Install the source checkout¶

These changes have not yet been published to PyPI. From the repository root:

python -m pip install -e "."

No ToolUniverse initialization is required. Scientific programs, model weights, reference data and API credentials remain task-specific requirements. See third-party notices.

Optional genomics extensions¶

From this checkout, install .[genomics-genes] for Pyrodigal, .[dna-storage] for Chamaeleo, or .[dna-nanotechnology] for scadnano. Breeding uses .[bio-analysis]; the MCP client uses .[agent].

Install the local gene-finding extension:

python -m pip install -e ".[genomics-genes]"

External executables, references and model weights remain separate; see external genomics and MCP.