Integrations

Storage

Read & write:

• Local file systems

• AWS S3

• S3-compatible storage services, e.g., MinIO & Cloudflare R2

• Google Cloud Storage

Read only:

• http/https: ln.Artifact("https://a-website.com/a-dataset.txt")

• Hugging Face: ln.Artifact("hf://datasets/org/repo/dataset.parquet")

Ontologies & registries

• bionty: Biological ontologies, with easy import from >20 public ontologies

• pertdb: Registries for perturbations (compounds, biologics, genetic interventions, etc.)

Git

Auto-sync with git: track guide.

Computational notebooks

You can call ln.track() and ln.finish() (or the R versions) in four notebook types:

• Jupyter

• Marimo

• R markdown

• Quarto

For both notebook types, source code is stored in .py format and run reports as html. Jupyter notebooks leverage the py:percent format and Marimo notebooks the Marimo standard.

MLOps

• PyTorch Lightning: lightning

• Weights & Biases: see the Weigths & Biases guide

• MLFlow: see the MLFlow guide

• Croissant format: see the Croissant guide

• scVI: see the guide on the scVI docs

Workflow managers

• nextflow & the Seqera platform: see the Nextflow guide and the nf-lamin reference

• redun: via the Python API, see the redun guide

• prefect, airflow, dagster, etc.: see the workflows guide

• snakemake: via post-run logic, see the snakemake example

Tables & arrays

• pyarrow & polars: see the engine argument of open() and the arrays guide

• tiledbsoma: inhouse guide or cellxgene

• duckdb: via parquet files, see rxrx

Visualization

• Vitessce: see the vitessce guide

• DataVisyn: reach out for custom visualization solutions

ELN systems

• Benchling: sync schema and data from your registries within our team/enterprise plan