bloom-lims 2.0.2

Templated Abstract Polymorphic LIMS - A Laboratory Information Management System

Bloom: Templated Abstract Polymorphic (and opinionated) LIMS

a conceptual gambit in collaboration with chatGPT4 · v2.0.0

Built from first principles drawing on 30 years of scaling laboratory process. Constructed with as few object-model shortcuts as possible — because those shortcuts are among the main reasons LIMS nearly universally disappoint. Bloom supports arbitrary and prescribed interacting objects and is intended for small-to-factory-scale laboratories, regulated environments, and both research and operational use cases. It covers accessioning, lab processes, specimen/sample management, equipment, and regulatory compliance.

What Bloom Owns

Bloom is the wet-lab and material-state authority:

• containers and placements
• specimens and derived materials
• extraction, QC, library-prep, pool, and run objects
• sequenced library assignments
• wet-lab queue membership and queue-transition state
• Bloom-side lineage linking to Atlas fulfillment context

What Bloom Does Not Own

• customer-portal data and tenant administration
• patient, clinician, shipment, TRF, or test truth
• artifact registry authority
• analysis execution or result-return workflows

Spoilers

bloom early peeks

AWS Cognito Authentication

with flexible whitelisting, role-based access, and session management

• Bloom Cognito configuration docs

Graph Object View (add, remove, edit, take actions, explore)

Interactive, Dynamic Metrics

Accessioning Modalities

Nested Assay / Queue / Workset

Instantiate Objects From Available Templates

Object Detail

Specialized Object Detail Views

Labware (ie: a 96w plate)

Bloom natively supports arbitrarily defined labware — a 96w plate is one example. Anything describable as nested arrays of arrays can be configured as labware with next to no code.

Exhaustive & Comprehensive Audit Trails (+soft deletes only)

Bells And Whistles

• Integrated with FedEx tracking for entered FedEx barcodes

Integrated Barcode Label Printing For All Objects

• See detailed docs here

  

  • Leverages the zebra_day library

Workflows Available

Accessioning

Package receipt → kit registration → specimen registration → requisition capture & association → adding specimens to assay queues. FedEx tracking fetched, barcode printing available.

Plasma Isolation → DNA Extraction → DNA Quant

Manages all object relationships, tracks all details, prints labels.

Installation

Prerequisites

Hardware

• macOS 14+ (intel or Apple Silicon)
  • brew install coreutils required for gtimeout (rclone). Add alias timeout=gtimeout to your shell config.
• Ubuntu 22+

Conda

Bloom requires Conda for environment management. Install Miniconda:

Linux x86_64:

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
~/miniconda3/bin/conda init && bash

macOS:

# Intel
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
# Apple Silicon
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-arm64.sh
bash Miniconda3-latest-MacOSX-*.sh
~/miniconda3/bin/conda init && bash

AWS Cognito

Authentication is handled via AWS Cognito. Follow the Cognito configuration guide before starting the server.

Quick Start

Assumes Conda is installed and Cognito is configured.

# Clone and enter the repository
git clone [e-mail]:Daylily-Informatics/bloom.git
cd bloom

# Activate environment (creates the BLOOM conda env on first run)
source bloom_activate.sh

# Initialize and seed the database
bloom db init
bloom db seed

# Start the web UI (default: http://localhost:8911)
bloom gui

Optional: pgAdmin4 Database Admin UI

source bloom_lims/env/install_pgadmin.sh

Optional: Install via PyPI

pip install bloom_lims

Runtime Shape

Bloom is a FastAPI application with both API and server-rendered GUI surfaces.

Primary entrypoints:

• App entrypoint: main.py (run via uvicorn main:app)
• App factory: bloom_lims.app:create_app
• CLI entrypoint: bloom (after source bloom_activate.sh)

Primary CLI groups:

Command	Description
bloom db	init, start, stop, status, migrate, seed, shell, reset
bloom gui	Start the BLOOM web UI
bloom server	start, stop, status, logs
bloom config	Show or edit configuration
bloom info	Show environment information
bloom status	Check service status
bloom doctor	Verify environment health
bloom shell	Interactive Python shell
bloom logs	View service logs
bloom integrations	Atlas integration management
bloom quality	Code quality checks
bloom test	Run targeted test suites
bloom users	User management

API Surface

Canonical prefix: /api/v1

Route group	Purpose
/api/v1/objects	Core LIMS objects
/api/v1/containers	Labware and placements
/api/v1/content	Specimen / material content
/api/v1/equipment	Equipment registry
/api/v1/templates	Object template management
/api/v1/subjects	Subject (patient/donor) records
/api/v1/lineages	Object lineage graph
/api/v1/search	Search v1
/api/v1/search/v2	Search v2
/api/v1/object-creation	Batch object creation
/api/v1/tracking	Carrier tracking
/api/v1/user-tokens	User token management
/api/v1/admin/*	Admin endpoints
/api/v1/external/specimens	External specimen intake
/api/v1/external/atlas	Atlas integration bridge
/api/v1/external/atlas/beta	Atlas beta endpoints

Server-rendered GUI routes remain active under the root app (login, operational views, graph screens).

Cross-Repo Boundaries

• Atlas sends accepted-material and status traffic into Bloom.
• Bloom links physical execution state back to Atlas TRF/test/fulfillment-item context.
• Bloom can register run artifacts in Dewey when Dewey integration is enabled.
• Ursa resolves sequencing context from Bloom using run and lane identifiers.

TapDB Mount

Bloom can mount the TapDB admin surface inside the same server process at /admin/tapdb.

Rules when mounted:

• Bloom session auth is the gate
• Access is admin-only
• Unauthenticated browser requests redirect to /login
• Mounted TapDB local login is disabled

Integrations

CRMs

If they have APIs, fetching physician or patient identifiers/metadata is straightforward.

Zebra Barcode Label Printing

• In place. See detailed docs here.
• Uses zebra_day.

FedEx Tracking API

• In place. Requires the config YAML for fedex_tracking_day.

Salesforce (example CRM)

• simple_salesforce or salesforce python packages — straightforward to add.

Design Principles

Enterprise UIDs (EUIDs)

Each Object Has A UUID; UUIDs Are Immutable And Not Reused

Using the same UUID on child objects for convenience creates irreconcilable confusion at scale.

The UID Identifies The Object Class; The UUID Identifies The Instance

Reference: don't put metadata in a UUID.

Exhaustive Object Metadata Is Queryable Via The Enterprise UUID

Metadata may also appear on printed barcode labels alongside the EUID.

Bloom EUIDs are uppercase-prefix + integer — safe as filenames across case-sensitive and case-insensitive file systems.

Trust The Database To Manage UUIDs

Clear And Concise Data Model

TSVs, Not CSVs

Few compelling reasons to use CSV over TSV; many reasons not to.

All LIMS Data Editable via CRUD UI

Fully editable — soft deletes only, ensuring complete audit coverage.

Object Definitions and Actions Are Config-Driven

Minimal code changes required to add new object types or workflow steps.

Other

• Simple · Scalable · Secure · Flexible & Extensible · Open Source · Operationally Robust · Free
• Sustainable (per the Snakemake rolling paper definition)

Use Cases

LIMS Actions (must have)

Many-To-Many Relationships Among All Objects

All other relationships are subsets of this. Blocking many-to-many leads to inflexibility.

Objects May Be Involved In Multiple Workflows Simultaneously

Support For Predefined and Arbitrary Workflows

Objects may be: root, child, parent, or terminal — composable in any combination

Zero Loss Of Data (comprehensive audit trails, soft deletes) && 100% Audit Coverage

Deployment & Maintenance

Bloom deploys wherever it runs. You own security, backups, recovery, performance, and monitoring. Consulting available.

Regulatory & Compliance

CLIA

No reason Bloom cannot be used in a CLIA regulated environment.

CAP

Bloom can satisfy all relevant CAP checklist items. Most items concern the environment Bloom is operated in.

HIPAA

If installed in a HIPAA-compliant environment, Bloom should require minimal additional work to comply.

Timezone Policy

• Bloom persists timestamps in UTC.
• Display timezone is user-configurable via TapDB-backed system_user preferences.
• Canonical preference key: display_timezone.

Support

No promises — please file issues for bugs or feature requests.

Authors

• John Major aka iamh2o
• Josh Durham
• Adam Tracy

License

MIT

References & Acknowledgments

• chatGPT4 — helped build this.
• Everyone who ran early versions and offered feedback.
• Snakemake — inspiration.
• MultiQC — inspiration.
• GA4GH — inspiration.
• The Human Genome Project — where I learned I loved LIS.
• Cytoscape — graph visualization.
• Semantic MediaWiki — inspiration.
• Datomic — inspiration.
• The OSS world.

Testing

source bloom_activate.sh
bloom db init
pytest

Focused validation (matches CI):

pytest --no-cov \
  tests/test_config_runtime.py \
  tests/test_api_atlas_bridge.py \
  tests/test_atlas_lookup_resilience.py \
  tests/test_queue_flow.py \
  tests/test_run_resolver.py

ruff check bloom_lims tests

Dev Tools

All commands assume source bloom_activate.sh has been run.

Reset and Rebuild the Database (⚠️ destroys all data)

bloom db reset
bloom db init
bloom db seed

Start the UI

# Via CLI (recommended)
bloom gui

# Or directly via uvicorn (dev mode, port 8911)
uvicorn main:app --reload --port 8911

# Or via gunicorn (production)
bash run_bloomui.sh --mode prod --port 8911

Interactive Python Shell

bloom shell

pgAdmin UI

source bloom_lims/env/install_pgadmin.sh

Lint / Format

ruff check bloom_lims tests
ruff format bloom_lims tests

Notes

File System Case Sensitivity

macOS Is NOT Case Sensitive

echo "test" > test.log && echo "TEST" > TEST.LOG
more test.log   # → TEST  (same file!)

Ubuntu Is Case Sensitive

echo "test" > test.log && echo "TEST" > TEST.LOG
more test.log   # → test

Assume Case Insensitivity In All File Names

Given files may be reconstituted on a case-insensitive file system, all file names should be treated as case-insensitive.

Bloom UUIDs and EUIDs Are Safe As File Names

Per RFC 4122, UUID uppercase/lowercase are equivalent. Bloom EUIDs use an uppercase prefix followed by integers only.