Lamin

Lamin is an open data framework for biology, combining LaminDB and LaminHub into a unified platform that provides context and memory for datasets and models at scale. It is designed for life sciences teams and AI agents that need to query, trace, and validate biological data through a lineage-native lakehouse supporting bio-formats, registries, and ontologies. Lamin is available as a Python package (pip install lamindb) and is built to serve both individual researchers and large collaborative organizations.

The platform addresses core challenges in biological data management: tracking where data came from, enforcing metadata consistency, unifying experimental records with stored datasets, and building an organization's long-term institutional memory — all without vendor lock-in.

Data Lineage Tracking

• Track data lineage with a single line of code across notebooks, scripts, pipelines, and functions.
• Know where data came from and what it is used for at every stage of the workflow.
• Supports Python, R, shell, Nextflow, and SQL environments.
• Trace data, code, and reports in a unified lineage graph.

Lakehouse and Bio-Format Support

• Query and batch-load datasets at scale using a lakehouse architecture.
• Supports a wide range of table and array formats including .parquet, .zarr, AnnData, and SpatialData.
• Manage dataset features and schemas as metadata stored in Postgres or SQLite.
• Designed to handle the specific data structures common in biological and spatial omics research.

Registries, Sheets, and LIMS Capabilities

• Manage metadata in relational sheets that remain in sync with datasets in storage.
• Use a single Python or R class with built-in ontologies, project management, and change management.
• Track experiments, samples, notes, reports, and more within a unified interface.
• Integrates ontologies directly into the metadata management workflow.

Data Integrity and Validation

• Use schemas to enforce consistency across all data assets.
• Annotate datasets with a single line of code.
• Supports validation for sheets, .parquet, .csv, .zarr, AnnData, SpatialData, and other formats.
• Ensures that data entering the lakehouse meets defined quality and structural standards.

Administration and Zero Lock-In

• Manage fine-grained permissions for both human users and AI agents with SaaS-like simplicity.
• Permissions are enforced directly at the database and storage level, preserving admin control.
• Supports deployment on AWS S3, Google Cloud Platform, Azure, Cloudflare R2, local file systems, and on-premises infrastructure.
• Compatible with both Postgres and SQLite as the metadata backend.

Organizational Memory and AI Readiness

• As teams and agents work, data, models, and reports are automatically mapped into the lakehouse.
• Builds recursively queryable memory and training data that compounds in value over time.
• Enables long-term institutional knowledge retention across biological research programs.

Lamin is deployable across major cloud providers and on-premises infrastructure, making it suitable for organizations that require data sovereignty alongside collaborative scale. Its open framework design ensures teams retain full administrative control while benefiting from structured, ontology-aware data management purpose-built for biology.