Feature List and Roadmap

This document outlines the features of the {laminr} package and the roadmap for future development.

Features

Connect to an instance

• Connect to a LaminDB instance (connect()).
• Handle authentication and authorization.
• Connect to a LaminDB instance without needing to install the lamin_cli Python package.

Query & search

• Query exactly one record (Registry$get(...)): Fetch a single record by ID.
• Query sets of records (Registry$filter()): Fetch multiple records based on filters.
  • $df(): Returns a data frame with each record in a row.
  • $all(): Returns all records as a QuerySet.
  • $one(): Return exactly one record.
  • $one_or_none(): Return one record or NULL.
• Leverage relationships when querying (Artifact$filter(created_by__handle__startswith = "testuse")$df()): Query records based on relationships.
• Comparators: Use comparators in filters.
  • and: Example: Artifact$filter(suffix = ".jpg", created_by = user)
  • less than / greater than: Example: Artifact$filter(size__lt = 1e4)
  • in: Example: Artifact$filter(suffix_in = [".jpg", ".fastq.gz"])
  • order by: Example: Artifact$filter().order_by("created_at")
  • contains: Example: Artifact$filter(name__contains = "test")
  • startswith: Example: Artifact$filter(name__startswith = "test")
  • or: Example: ...
  • not: Example: ...
• Search for records (Registry$search(...)): Search for records based on a query string.
• Pagination: Support pagination for large query results.
• Field lookups: Provide convenient functions for looking up field values (e.g., Artifact$lookup("description")).

Manage data & metadata

• Create artifacts: Create new artifacts from various data sources (e.g., files, data frames, in-memory objects).
  • $from_df(): Create an artifact from a data frame.
  • $from_path(): Create an artifact from a path.
  • $from_anndata(): Create an artifact from an AnnData.
• Save artifacts: Save artifacts to LaminDB with appropriate metadata.
• Load artifacts: Load artifacts from LaminDB into R:
  • csv: Load a data frame from a CSV file.
  • fcs: Load flow cytometry data.
  • h5ad: Load an AnnData from an HDF5 file.
  • h5mu: Load a MuData from an HDF5 file.
  • html: Load content from an HTML file.
  • jpg: Load an image from JPG.
  • json: Load data from a JSON file.
  • parquet: Load a data frame from a Parquet file.
  • png: Load an image from PNG.
  • rds: Load an R object from an RDS file.
  • svg: Load an image from SVG.
  • tsv: Load a data frame from a TSV file.
  • yaml: Load data from a YAML file.
  • zarr: Load an AnnData from a Zarr store.
• Cache artifacts: Cache artifacts locally for faster access:
  • s3: Interact with S3 storage.
  • gcp: Interact with Google Cloud Storage.
• Version artifacts: Create new versions of artifacts.
• Delete artifacts: Delete an existing artifact.
• Manage artifact metadata: Add, update, and delete artifact metadata.
• Work with collections: Create, manage, and query collections of artifacts.

Track notebooks & scripts

• Track code execution: Automatically track the execution of R scripts and notebooks.
• Capture run context: Record information about the execution environment (e.g., package versions, parameters).
• Link code to artifacts: Associate code execution with generated artifacts.
• Visualize data lineage: Create visualizations of data lineage and dependencies.
• Finalize tracking: End and save a run.

Curate datasets

• Validate data: Validate data against predefined schemas or constraints.
• Standardize data: Apply standardization rules to ensure data consistency.
• Annotate data: Add annotations and labels to data.
• Use the Curator class: Implement the Curator class for a streamlined curation workflow.

Access public ontologies

• Access ontology data: Fetch data from public ontologies (e.g., gene names, protein IDs).
• Search ontologies: Search for entities within ontologies.
• Use ontology terms in queries: Use ontology terms to filter and query data.
• Manage ontology versions: Access different versions of ontologies.

Manage biological registries

• Create and manage records in bionty registries: Add, update, and delete records for genes, proteins, cell types, etc.
• Utilize hierarchical relationships: Navigate and query based on parent-child relationships in ontologies.
• Manage synonyms: Add and use synonyms for biological entities.

Manage schema modules

• List available modules: Retrieve a list of available modules in an instance.
• Access module registries: Access registries within specific modules.
• (Advanced) Create custom modules: Define and register custom schema modules.

Transfer data

• Upload data: Upload data files to LaminDB storage.
• Download data: Download data files from LaminDB storage.
• (Advanced) Support zero-copy data transfer: Implement efficient data transfer mechanisms.

Roadmap

Version 0.1.0

A first version of the package that allows users to:

• Connect to a LaminDB instance.
• List all records in a registry.
• Fetch one record by ID or UID.
• Cache S3 artifacts locally.
• Load AnnData artifacts.

Version 0.2.0

• Expand query functionality with comparators, relationships, and pagination.
• Implement basic data and metadata management features (create, save, load and delete artifacts).
• Expand support for different data formats and storage backends.
• Implement code tracking.

Version 0.3.0

• Implement data lineage visualization.
• Introduce data curation features (validation, standardization, annotation).
• Enhance support for bionty registries and ontology interactions.

Future versions

• Implement advanced features like custom module creation and zero-copy data transfer.
• Continuously improve performance, usability, and documentation.