Changelog 2026 .md

Note

Get notified by watching releases for git repositories: lamindb, laminhub-public, laminr, nf-lamin, and bionty.

For other years, see: 2025 · 2024 · 2023 · 2022

2026-02-09 db 2.1.2

UX:

  • 🚸 Broaden curation to dtypes that are unions PR @Zethson

  • 🚸 More debug logging in case instance creation on hub fails due to quota exceeded PR @Koncopd

Performance:

  • ⚡️ Download large single files from s3 with boto3 by default in artifact.cache() PR @Koncopd

Bug fixes:

  • 🐛 Fix initializing an instance with a cloud storage and the db parameter passed PR @Koncopd

2026-02-09 hub 1.14.0

2026-02-02 db 2.1.1

UX:

  • 🚸 Better detection of read-only connections for write errors and tracking PR @Koncopd

  • 🚸 A lamin settings cache-dir reset CLI command for resetting the cache directory to the default path PR @Koncopd

  • 🚸 A S3QueryPath.is_bucket_versioned() method to check if the bucket in an s3 path is versioned PR @Koncopd

Bug fixes:

  • 🐛 Fix edge case in Django reset during ln.setup.init(...) PR @Koncopd

2026-02-03 nf 0.5.0

This release revamps the configuration (nextflow.config) for more control over which artifacts are registered and which metadata is attached (kind, projects, ulabels). Here is an example Nextflow config:

lamin {
  instance = "laminlabs/lamindata"
  api_key = secrets.LAMIN_API_KEY
  output_artifacts {
    rules {
      output {
        pattern = ".*"
        order = 1
        type = "include"
        project_uids = ["AbCdEfGhIjKl"]
        ulabel_uids = ["AbCd0123"]
      }
      // don't publish state.yaml files
      yaml {
        pattern = ".*\\.state\\.yaml"
        type = "exclude"
        order = 10
      }
    }
  }
}

New functionality:

  • ✨ Control which files are registered as artifacts using pattern-based rules PR @rcannood

  • ✨ Link projects and ulabels to transforms, runs, and artifacts with new run and transform config sections for granular control PR @rcannood

  • ✨ New api section for connection settings replaces root-level fields PR @rcannood

UX:

  • 🚸 Better mapping of HTTP error codes from the LaminDB API PR @rcannood

Bug fixes:

  • 🐛 Fixed edge case in tracking input files and improved S3 path rendering PR @rcannood

Deprecations: The following API components are now deprecated and are scheduled for removal in v0.6

Type

Deprecated

Replacement

Config

project

project_uids

Config

supabase_api_url, supabase_anon_key, max_retries, retry_delay

api { ... } section

Function

getArtifactFromUid()

file('lamin://owner/instance/artifact/uid')

2026-02-02 db 2.1.0 | bionty 2.1.0

MLOps:

Notebooks, scripts, workflows:

  • 🚸 Recognize packaged python source code across track(), @flow(), and @step() PR @falexwolf

  • 🚸 Default to creating a global run context in ln.flow() PR @falexwolf

  • 🚸 Serialize Path, UPath, and handle None in transform params PR @falexwolf

  • 🚸 Reflect user CLI calls in transform description PR @falexwolf

  • 🚸 Improve error message for running a script or notebook outside the development directory PR @falexwolf

Validation:

  • 🚸 Unify validating features with and without the presence of a schema, fix parsing ISO-format strings as datetime if mapped on a feature with dtype=str PR @falexwolf

CLI:

  • 🚸 Delete records via CLI PR @falexwolf

  • 🚸 Better error behavior in lamin save when transform is on a different branch PR @sunnyosun

General UX:

  • 🚸 Enable convenient re-upload of local file after upload failure PR @falexwolf

  • 🚸 Return information about deleted objects rather than None upon hard delete PR @falexwolf

  • 🚸 Introduce safe-guards in updating the space of artifacts PR @Koncopd

  • 🚸 Switch on SQL logging if ln.settings.verbosity is "debug" PR @sunnyosun

  • 🚸 Raise canonical InstanceNotFoundError in Registry.connect() PR @falexwolf

  • 🚸 Allow users to pass ids of foreign keys in Artifact() constructor PR @Koncopd

  • 🚸 Check local path existence on artifact.save() PR @Koncopd

Performance:

  • ⚡️ Much faster permanent run & transform deletion PR @falexwolf

  • ⚡️ 1 sec speed-up for basic CLI calls through optimized lamindb_setup import time PR @falexwolf

Ontology updates in bionty:

Better conventions in CLI (backward compatible):

  • 🚚 Support key-based deletion in lamin delete PR @falexwolf

  • 🚚 lamin cache is now lamin settings cache-dir, refactor settings managemennt PR @falexwolf

  • 🚚 More consistent lamin switch and lamin create commands PR @falexwolf

The wetlab plugin was renamed to pertdb (backward compatible):

Bugs:

  • 🐛 Refactor & fix feature-based queries for records PR @falexwolf

  • 🐛 Fix PyYAML serialization of NotebookNode PR @Zethson

  • 🐛 Fix Artifact.from_dataframe() to support path input PR @Zethson

  • 🐛 Fix standardize() ignoring passed source PR @Zethson

Docs:

  • 📝 Convert guides to markdown format PR @falexwolf

  • 📝 Legible API reference for lamindb.models PR @falexwolf

  • 📝 Consistent reference to Python/R alternative for all CLI commands, expand lamin annotate PR @falexwolf

  • 📝 Document backward relationships and add RelatedManager[T] to ManyToMany type annotations PR PR @falexwolf

  • 📝 README improvements PR @falexwolf

Refactoring:

  • ♻️ Rather flag _storage_ongoing than _storage_completed PR @falexwolf

  • ♻️ Efficiently encode uploading information PR @falexwolf

  • ♻️ Fully leverage Context to manage the run context of function executions PR @falexwolf

  • ♻️ Use publishable LaminHub key PR @Koncopd

  • ♻️ Remove internal use_proxy_db option PR @Koncopd

2026-01-31 hub 1.13.0

  • ⚡ Update JWT signing from HS256 to ES256 for better performance PR @fredericenard

  • 💄 Display output and input records on run detail page PR @falexwolf

2026-01-27 hub 1.12.1

2026-01-26 hub 1.12.0

  • ✨ Configurable default Seqera workspace and compute environment PR @chaichontat

  • ✨ Two new endpoints to help users authenticate and interact with instances through the API PR @fredericenard

  • 🐛 Fix attaching a space to an instance in case an org admin isn’t a collaborator of that space PR @fredericenard

  • 🐛 Prevent empty results when searching Artifacts PR @chaichontat

  • 💄 Landing page: re-format taglines and update schematic PR @chaichontat

2026-01-21 nf 0.4.0

✨ This release allows for the automatic tracking of input artifacts by discovering their remote URIs, using functionality implemented in Nextflow 25.10. Hence, it requires Nextflow >=25.10.0 PR @rcannood

export NXF_VER=25.10.2
nextflow run <pipeline> \
    --input "https://raw.githubusercontent.com/..." \
    -with-report report.html

2026-01-20 hub 1.11.0

2026-01-14 hub 1.10.0

  • ✨ Add /artifacts/by-path endpoint PR @fredericenard

  • 🐛 Write run annotations on the Launch page into the RunRecord registry PR @falexwolf

  • 📝 Improve landing page - metadata and lakehouse sections PR @falexwolf

2026-01-13 hub 1.9.1

  • 🐛 More strictly enforce schema during UI-based record creation PR @chaichontat

  • 🐛 Broaden database server access permissions to include shared database servers PR @fredericenard

  • 📝 Improve landing page - introduce tags and improve language PR @falexwolf

2026-01-13 R 1.2.2

2026-01-12 nf 0.3.0

  • ✨ Implement FileSystem for lamin:// URIs PR @rcannood

  • ⬆️ Bump lamindb to v2 PR @falexwolf

  • 🚸 Update formatting of source_code when a new transform is created PR @rcannood

  • 📝 Add version compatibility table PR @rcannood

2026-01-12 hub 1.9.0

  • ✨ Transform search bar also understands run names and UIDs PR @chaichontat

  • 🚸 Default to not listing inferred schemas without names PR @chaichontat

  • 🐛 Fix the broken “New schema” button on the records page due to lamindb v2 PR @chaichontat

  • 🐛 Fix the broken pipeline Launch button due to lamindb v2 PR @sunnyosun

2026-01-11 db 2.0.1

  • 🚸 Backward compatibility for querying features by dtype PR @falexwolf

2026-01-11 db 2.0.0

This release makes the database schema fully consistent and thereby improves UX, integrity, and enables long-term stability.

⚠️ With the exception of previous deprecations, your code continues to run as is, but you will receive new warnings about renamed API components.

Main changes:

What

Before

After

Backward compatible

Function tracking

@ln.tracked()

@ln.flow(), @ln.step()

✓

Feature field

dtype

_dtype_str

✓

Schema field

n

n_members

✓

IsVersioned field

version

version_tag

✓

Transform field

type

kind

✓

Artifact field

feature_sets

schemas

✓

Feature field

coerce_dtype

coerce

✓

Schema field

coerce_dtype

coerce

✓

FeatureValue registry

FeatureValue

JsonValue

✓

Private fields that are now public:

What

Before

After

Backward compatible

Artifact field

_subsequent_runs

recreating_runs

✗

Artifact field

_feature_values

json_values

✗

Run field

_recreated_artifacts

recreated_artifacts

✗

SQLRecord field

_branch_code (only on DB-level)

branch_id

✗

Changes to auxiliary (JSON) fields:

What

Before

After

Backward compatible

Run field

cli_args (in _aux)

cli_args (CharField(max_length=1024))

✓

Feature field

default_value (in _aux)

default_value (JSONField)

✓

Feature field

nullable (in _aux)

nullable (BooleanField)

✓

Schema field

flexible (in _aux)

flexible (BooleanField)

✓

Artifact field

_is_saved_to_storage_location (in _aux)

_save_completed (in _aux)

✗

On link models with labels:

What

Before

After

Backward compatible

IsLink model

feature_ref_is_name

removed

✗

IsLink model

label_ref_is_name

removed

✗

For completeness, changes in recently introduced unused API components:

What

Before

After

Backward compatible

RootBlock registry

RootBlock

Block

✗

Block field

context

key

✗

BaseBlock field

type

kind

✗

Run field

_logfile

removed

✗

Transform field

config

removed

✗

Transform field

flow

removed

✗

Transform field

is_flow

removed

✗

Transform field

_template

removed

✗

Changes in the bionty and wetlab modules:

What

Before

After

Backward compatible

BioRecord registry

uid (X chars)

uid (14 chars)

* .get() of the X chars uid continues to work

* uids are re-encoded to 14 chars with the first X chars conserved.

Changes in the wetlab module:

What

Before

After

Backward compatible

Experiment registry

Experiment

removed

✗

BioSample registry

BioSample

removed

✗

TechSample registry

TechSample

removed

✗

Donor registry

Donor

removed

✗

Well registry

Well

removed

✗

GeneticPerturbation field

system

type

✗

CompoundPerturbation field

targets

removed

✗

Compound field

targets

✓

If you have data in removed registries you’ll see a warning that your data gets exported to parquet files. You can import these parquet files into the Record registry under an Experiment type without losing of functionality.

⚠️ After migrating databases to v2 via lamin migrate deploy lamindb package installations with v1 will raise errors. This is the last planned database migration that breaks older clients.

All instances connected to LaminHub have been migrated and there is no need to act.

If you are an admin of a self-managed instance, please migrate your database with lamin migrate deploy.

For details see the Compatibility Matrix and the source code.

⚠️ For laminr and nf-lamin users

For laminr users with an auto-managed reticulate environment, please run this:

reticulate::py_require("lamindb>=2.0.1") # DO THIS FIRST
ln <- laminr::import_module("lamindb")

For nf-lamin users, please upgrade to 0.3.0.

Removals:

Features:

  • ✨ Track shell scripts PR @falexwolf

  • ✨ Enable managing workflows by decorating functions with @ln.flow() and @ln.step() and deprecate @ln.tracked() PR @falexwolf

Architecture:

  • 🏗️ For more robustness and to convey the field’s special nature in the API and the database schema, replace Feature.dtype with Feature._dtype_str to now hold a serialization that links object types (Record + ULabel) based on their uid instead of their name and type hierarchy PR @falexwolf

UX:

  • 🚸 Add feature.dtype_as_object PR @sunnyosun

  • 🚸 Add description argument to Transform.from_git() PR @falexwolf

  • 🚸 Allow passing _key_is_virtual on Artifact creation PR @Koncopd

  • 🚸 Raise SystemExit upon InstanceNotFound only from CLI PR @falexwolf

  • 🚸 Raise error for init or import issues in lamin io snapshot PR @Zethson

  • 🚸 Do not track inputs when calling collection.artifacts.all() PR @Koncopd

  • 🚸 Raise a clear error if calling CanCurate and similar registry class methods on objects PR @falexwolf

  • 🚸 Sort features in describe() PR @falexwolf

Database-level changes:

  • 🗃️ Rename RootBlock to Block with a column key akin to Artifact and Transform PR @falexwolf

  • 🗃️ Refactor _storage_completed mechanism and remove unique constraints on type names PR @falexwolf

  • 🗃️ Migrate most auxiliary fields to SQL columns PR @falexwolf

  • 🗃️ Simplify Transform PR @falexwolf

  • 🗃️ Rename IsVersioned.version to IsVersioned.version_tag, Transform.type to Transform.kind, FeatureValue to JsonValue, BaseBlock.type to BaseBlock.kind, delete Transform._template and Run._logfile PR @falexwolf

  • 🗃️ Rename Artifact._subsequent_runs to Artifact.recreating_runs and Run._recreated_artifacts to Run.recreated_artifacts PR @falexwolf

  • 🗃️ Rename SQLRecord._branch_code to SQLRecord.branch_id and remove feature_ref_is_name and label_ref_is_name fields from link models PR @falexwolf

  • 🗃️ Rename Artifact.feature_sets to Artifact.schemas PR @falexwolf

  • 🗃️ Migrate the Feature and Schema database tables to LaminDB v2 PR @falexwolf

  • 🗃️ Swap accessor names for SchemaComponent links, consistent ManyToMany field names for RecordRecord links PR @falexwolf

  • 🗃️ Better names for the ManyToMany fields for the annotating runs and artifacts of an Artifact PR @falexwolf

Bug fixes:

Bionty:

Wetlab:

  • 🗃️ Migrations for wetlab v2 PR @sunnyosun

  • 🗃️ Migrations for LaminDB v2 PR @falexwolf

  • 🗃️ Create a composite index for artifact, perturbationtarget fields in ArtifactPerturbationTarget PR @Koncopd

  • 🗃️ Create a composite index for artifact, geneticperturbation fields in ArtifactGeneticPerturbation PR @Koncopd

Docs:

  • 📝 Improve the Space document PR @falexwolf

  • 📝 Improve the Artifact.replace() docs PR @falexwolf

  • 📝 Improve FAQ doc on trashing and archiving PR @falexwolf

  • 📝 Render return type annotations of Artifact.open() in docs PR @falexwolf

Config:

Refactoring:

  • ♻️ Refactor require_organism PR @sunnyosun

  • ♻️ Better re-use Django’s ObjectDoesNotExist and MultipleObjectsReturned PR @falexwolf

  • ♻️ Return file size in hash_file() PR @falexwolf

  • ♻️ Squash migrations and re-build SQLite init database PR @falexwolf

  • ♻️ More verbose migration integrity check PR @falexwolf

  • ♻️ Call lamin migrate deploy even if initializing from pre-built sqlite file PR @falexwolf

Dependencies:

  • ⬆️ Bump fsspec packages upper bound PR @Koncopd

  • ⬆️ Bump anndata upper bound PR @Koncopd

  • ⬆️ Set the correct lower bound for websockets PR @Koncopd

2026-01-11 hub 1.8.0

2026-01-03 hub 1.7.0