madsci.resource_manager 0.4.6

The Modular Autonomous Discovery for Science (MADSci) Resource Manager.

MADSci Resource Manager

Tracks and manages the full lifecycle of laboratory resources - assets, consumables, samples, containers, and labware.

Features

• Comprehensive resource types: Assets, consumables, containers with specialized behaviors
• Complete history tracking: Full audit trail with restore capabilities
• Container hierarchy: Supports racks, plates, stacks, queues, grids, and custom containers
• Quantity management: Track consumable quantities with capacity limits
• Query system: Find resources by type, name, properties, or relationships
• Constraint validation: Prevents logical errors like negative quantities or overflow

Installation

See the main README for installation options. This package is available as:

• PyPI: pip install madsci.resource_manager
• Docker: Included in ghcr.io/ad-sdl/madsci
• Example configuration: See example_lab/managers/example_resource.manager.yaml

Dependencies: PostgreSQL database (see example_lab)

Usage

Quick Start

Use the example_lab as a starting point:

# Start with working example
docker compose up  # From repo root
# Resource Manager available at http://localhost:8003/docs

# Or run standalone
python -m madsci.resource_manager.resource_server

Manager Setup

For custom deployments, see example_resource.manager.yaml for configuration options.

Resource Client

Use ResourceClient to manage laboratory resources:

from madsci.client.resource_client import ResourceClient
from madsci.common.types.resource_types import Asset, Consumable, Grid
from madsci.common.types.resource_types.definitions import ResourceDefinition

client = ResourceClient("http://localhost:8003")

# Add a new asset (samples, labware, equipment)
sample = Asset(
    resource_name="Sample A1",
    resource_class="sample",
    attributes={"compound": "aspirin", "concentration": "10mM"}
)
added_sample = client.add_resource(sample)

# Add consumables with quantities
reagent = Consumable(
    resource_name="PBS Buffer",
    resource_class="reagent",
    quantity=500.0,
    attributes={"units": "mL"},
)
added_reagent = client.add_resource(reagent)

# Create containers (plates, racks, etc.)
plate = Grid(
    resource_name="96-well Plate #1",
    resource_class="plate",
    rows=8,
    columns=12
)
added_plate = client.add_resource(plate)

# Place samples in containers
client.set_child(resource=added_plate, key=(0, 0), child=added_sample)

# Query resources
samples = client.query_resource(resource_class="sample", multiple=True)
consumables = client.query_resource(resource_class="reagent", multiple=True)

# Manage consumable quantities
client.decrease_quantity(resource=added_reagent, amount=50.0)  # Use 50mL
client.increase_quantity(resource=added_reagent, amount=100.0) # Add 100mL

# Resource history and restoration
history = client.query_history(resource_id=added_sample.resource_id)
client.remove_resource(resource_id=added_sample.resource_id)  # Soft delete
client.restore_deleted_resource(resource_id=added_sample.resource_id)

Resource Types

Core Resource Hierarchy

Base Types:

• Resource: Base class for all resources
• Asset: Non-consumable resources (samples, labware, equipment)
• Consumable: Resources with quantities that can be consumed

Container Types:

• Container: Base for resources that hold other resources
• Collection: Supports random access by key
• Row: Single-dimensional containers
• Grid: Two-dimensional containers (plates, racks)
• VoxelGrid: Three-dimensional containers
• Slot: Holds exactly zero or one child (plate nests)
• Stack: LIFO access (stacked plates)
• Queue: FIFO access (sample queues)
• Pool: Mixed/collocated consumables

Usage Examples

# Different container types
tip_box = Grid(resource_name="Tip Box", rows=8, columns=12, resource_class="tips")
plate_stack = Stack(resource_name="Plate Stack", resource_class="plate_storage")
sample_rack = Row(resource_name="Sample Rack", length=24, resource_class="rack")

# Container operations
client.set_child(resource=tip_box, key=(0, 0), child=tip_sample)    # Grid access
client.push(resource=plate_stack, child=new_plate)                  # Stack push
client.pop(resource=plate_stack)                                    # Stack pop
client.set_child(resource=sample_rack, key=5, child=sample)         # Row access

Integration with MADSci Ecosystem

Resources integrate seamlessly with other MADSci components:

• Workflows: Reference resources in step locations and arguments
• Nodes: Access resource information during actions
• Data Manager: Link data to specific resources and samples
• Event Manager: Track resource lifecycle events

# Example: Node action using resources
@action
def process_sample(self, sample_resource_id: str) -> ActionResult:
    # Get sample attributes from Resource Manager
    sample = self.resource_client.get_resource(sample_resource_id)

    # Process based on sample properties
    result = self.device.analyze(sample.attributes["compound"])

    # Update sample with results
    sample.attributes["analysis_result"] = result
    self.resource_client.update_resource(sample)

    return ActionSucceeded(data=result)

Advanced Operations

Resource Definitions

Use ResourceDefinition for idempotent resource creation:

from madsci.common.types.resource_types.definitions import ResourceDefinition

# Creates new resource or attaches to existing one
resource_def = ResourceDefinition(
    resource_name="Standard Buffer",
    resource_class="reagent"
)
resource = client.init_resource(resource_def)  # Idempotent

Bulk Operations

# Query multiple resources
all_samples = client.query_resource(resource_class="sample", multiple=True)
empty_containers = client.query_resource(is_empty=True, multiple=True)

# Batch operations for consumables
for reagent in reagents:
    client.decrease_quantity(resource=reagent, amount=usage_amounts[reagent.resource_id])

History and Auditing

# Full resource history
history = client.query_history(resource_id=sample.resource_id)

# Query by time range and change type
import datetime
recent_updates = client.query_history(
    start_date=datetime.datetime.now() - datetime.timedelta(days=7),
    change_type="Updated"
)

Resource Templates

ResourceTemplates provide reusable blueprints for creating standardized laboratory resources. Templates help ensure consistency across resource creation and reduce configuration errors.

Creating Templates

Templates are created from existing resources and can be customized with metadata:

from madsci.client.resource_client import ResourceClient
from madsci.common.types.resource_types import Grid, Consumable

client = ResourceClient("http://localhost:8003")

# Create a standard 96-well plate resource
standard_plate = Grid(
    resource_name="Standard 96-Well Plate",
    resource_class="plate",
    rows=8,
    columns=12,
    attributes={
        "well_volume": 200,  # µL
        "material": "polystyrene",
        "sterilized": True
    }
)

# Create template from the resource
plate_template = client.create_template(
    resource=standard_plate,
    template_name="standard_96_well_plate",
    description="Standard 96-well polystyrene plate for assays",
    required_overrides=["resource_name"],  # Must be customized when using
    tags=["plate", "96-well", "assay", "standard"],
    created_by="lab_manager",
    version="1.0.0"
)

Using Templates to Create Resources

Templates streamline resource creation with consistent defaults:

# Create new resources from template
assay_plate_1 = client.create_resource_from_template(
    template_name="standard_96_well_plate",
    resource_name="Assay Plate #001",
    overrides={
        "attributes": {"experiment_id": "EXP001", "assay_type": "ELISA"}
    }
)

assay_plate_2 = client.create_resource_from_template(
    template_name="standard_96_well_plate",
    resource_name="Assay Plate #002",
    overrides={
        "attributes": {"experiment_id": "EXP002", "assay_type": "cell_culture"}
    }
)

# Both plates inherit standard configuration but with custom attributes

Template Management Operations

Listing and Discovery:

# List all available templates
all_templates = client.list_templates()

# Filter templates by category
plate_templates = client.list_templates(base_type="container", tags=["plate"])
reagent_templates = client.list_templates(base_type="consumable", tags=["reagent"])

# Get templates organized by category
templates_by_category = client.get_templates_by_category()
# Returns: {"container": ["plate_template", "rack_template"], "consumable": ["buffer_template"]}

# Filter by creator
lab_templates = client.list_templates(created_by="lab_manager")

Template Metadata:

# Get detailed template information
template_info = client.get_template_info("standard_96_well_plate")

# Returns metadata dictionary:
# {
#   "description": "Standard 96-well polystyrene plate for assays",
#   "required_overrides": ["resource_name"],
#   "tags": ["plate", "96-well", "assay", "standard"],
#   "created_by": "lab_manager",
#   "version": "1.0.0",
#   "created_at": "2024-01-15T10:30:00Z",
#   "resource": <template_resource_object>
# }

Template Updates:

# Update template metadata
updated_template = client.update_template(
    template_name="standard_96_well_plate",
    updates={
        "description": "Updated standard 96-well plate with new specifications",
        "tags": ["plate", "96-well", "assay", "standard", "v2"],
        "version": "1.1.0",
        "attributes": {"well_volume": 250}  # Updated well volume
    }
)

Template Deletion:

# Remove template (permanent)
success = client.delete_template("obsolete_template")
if success:
    print("Template successfully deleted")

Template Use Cases

1. Standardized Labware:

# Create templates for common labware
tip_box_template = client.create_template(
    resource=Grid(resource_name="Standard Tip Box", rows=8, columns=12, resource_class="tips"),
    template_name="standard_tip_box",
    description="200µL tip box template",
    required_overrides=["resource_name"],
    tags=["tips", "consumable", "standard"]
)

# Create multiple tip boxes from template
for i in range(5):
    client.create_resource_from_template(
        template_name="standard_tip_box",
        resource_name=f"Tip Box #{i+1:03d}",
        overrides={"attributes": {"batch_number": f"TB{i+1:03d}"}}
    )

2. Reagent Standards:

# Create reagent template
buffer_template = client.create_template(
    resource=Consumable(
        resource_name="PBS Buffer",
        resource_class="buffer",
        quantity=1000.0,
        attributes={"pH": 7.4, "concentration": "1X", "units": "mL"}
    ),
    template_name="pbs_buffer_1x",
    description="1X PBS buffer, pH 7.4",
    required_overrides=["resource_name", "quantity"],
    tags=["buffer", "pbs", "cell_culture"]
)

# Create buffer instances
buffer_stock = client.create_resource_from_template(
    template_name="pbs_buffer_1x",
    resource_name="PBS Stock #001",
    overrides={"quantity": 5000.0, "attributes": {"lot_number": "PBS2024001"}}
)

3. Container Hierarchies:

# Template for plate storage systems
storage_template = client.create_template(
    resource=Stack(resource_name="Plate Storage", resource_class="storage", capacity=20),
    template_name="plate_storage_stack",
    description="Standard plate storage stack (20 plates)",
    required_overrides=["resource_name"],
    tags=["storage", "plate", "stack"]
)

# Create storage locations
incubator_storage = client.create_resource_from_template(
    template_name="plate_storage_stack",
    resource_name="Incubator Plate Stack",
    overrides={"attributes": {"temperature": 37, "humidity": 95}}
)

Template Best Practices

1. Use Meaningful Names and Tags:

# ✅ Good - Descriptive and searchable
client.create_template(
    resource=plate,
    template_name="corning_96_well_flat_bottom",
    tags=["plate", "96-well", "flat-bottom", "corning", "cell-culture"]
)

# ❌ Avoid - Generic and hard to find
client.create_template(resource=plate, template_name="plate1", tags=["lab"])

2. Define Required Overrides:

# ✅ Good - Enforce customization of unique fields
client.create_template(
    resource=sample,
    template_name="dna_sample_template",
    required_overrides=["resource_name", "attributes.sample_id", "attributes.source"]
)

# ❌ Avoid - No required overrides may lead to duplicate names
client.create_template(resource=sample, template_name="sample_template")

3. Version Your Templates:

# Version templates for tracking changes
client.create_template(
    resource=updated_plate,
    template_name="assay_plate_v2",
    description="Updated assay plate with improved specifications",
    version="2.0.0",
    tags=["plate", "assay", "v2"]
)

4. Organize with Categories:

# Use consistent tag hierarchies
consumable_tags = ["consumable", "reagent", "buffer"]
labware_tags = ["labware", "plate", "96-well"]
equipment_tags = ["equipment", "analyzer", "hplc"]

Resource Locking and Concurrency Control

MADSci provides comprehensive resource locking to prevent conflicts when multiple processes or nodes access the same resources concurrently.

Basic Resource Locking

from madsci.client.resource_client import ResourceClient

client = ResourceClient("http://localhost:8003")

# Acquire lock on a single resource
success = client.acquire_lock(
    resource=sample_plate,
    lock_duration=300.0,  # 5 minutes
)

if success:
    try:
        # Perform operations on the locked resource
        client.set_child(resource=sample_plate, key=(0, 0), child=new_sample)
        client.update_resource(sample_plate)
    finally:
        # Always release the lock
        client.release_lock(resource=sample_plate)

Context Manager for Automatic Lock Management

The recommended approach uses context managers for automatic lock acquisition and release:

# Single resource locking
with client.lock(sample_plate) as locked_plate:
    # Resource is automatically locked
    locked_plate.set_child(key=(0, 0), child=new_sample)
    locked_plate.update_resource()
    # Lock automatically released when exiting context

# Multiple resource locking (atomic)
with client.lock(reagent_bottle, sample_rack, plate_stack) as (reagent, rack, stack):
    # All resources locked atomically or operation fails
    reagent.decrease_quantity(amount=50.0)
    new_sample = rack.get_child(key=5)
    stack.push(child=finished_plate)
    # All locks released automatically

Advanced Locking Patterns

Lock Duration and Auto-Refresh:

# Custom lock duration with auto-refresh
with client.lock(
    resource=long_running_plate,
    lock_duration=60.0,     # 1 minute initial lock
    auto_refresh=True,      # Automatically extend lock if needed
) as locked_plate:
    # Perform long-running operations
    # Lock automatically refreshed every 30 seconds
    for i in range(96):  # Process each well
        process_well(locked_plate, well_position=i)

Lock Status Checking:

# Check if resource is currently locked
is_locked = client.is_locked(resource=sample_plate)

if not is_locked:
    with client.lock(sample_plate) as locked_plate:
        perform_analysis(locked_plate)
else:
    print("Resource currently in use by another process")

Error Handling and Lock Recovery

# Manual lock management with error handling
try:
    if client.acquire_lock(resource=critical_resource, lock_duration=120.0):
        try:
            # Critical operations
            perform_critical_work(critical_resource)
        finally:
            client.release_lock(resource=critical_resource)
    else:
        raise Exception("Failed to acquire lock on critical resource")
except Exception as e:
    print(f"Operation failed: {e}")

Best Practices for Resource Locking

1. Always Use Context Managers:

# ✅ Good - Automatic cleanup
with client.lock(resource) as locked_resource:
    work_with_resource(locked_resource)

# ❌ Avoid - Manual management prone to errors
client.acquire_lock(resource)
work_with_resource(resource)
client.release_lock(resource)  # May not execute if exception occurs

2. Lock Multiple Resources Atomically:

# ✅ Good - All locks acquired or none
with client.lock(plate, reagent, tip_rack) as (p, r, t):
    transfer_samples(from_plate=p, reagent=r, tips=t)

# ❌ Avoid - Deadlock potential
with client.lock(plate) as p:
    with client.lock(reagent) as r:  # Could deadlock if another process locks in reverse order
        transfer_samples(p, r)

3. Use Appropriate Lock Durations:

# Short operations - brief locks
with client.lock(sample, lock_duration=30.0) as s:
    result = quick_measurement(s)

# Long operations - longer locks with auto-refresh
with client.lock(plate_stack, lock_duration=300.0, auto_refresh=True) as stack:
    process_entire_stack(stack)  # May take several minutes

Integration with Node Actions

Resource locking integrates seamlessly with MADSci node actions:

from madsci.node_module.node_module import RestNode
from madsci.common.types.action_types import ActionResult, ActionSucceeded

class AnalyzerNode(RestNode):

    @action
    def analyze_sample(
        self,
        sample_plate_id: str,
        sample_position: tuple[int, int]
    ) -> ActionResult:
        # Acquire lock before manipulating resources
        with self.resource_client.lock(sample_plate_id) as plate:
            # Get sample from locked plate
            sample = plate.get_child(key=sample_position)

            # Perform analysis
            result = self.instrument.analyze(sample)

            # Update sample with results
            sample.attributes["analysis_result"] = result
            plate.set_child(key=sample_position, child=sample)

            return ActionSucceeded(data=result)

Examples: See example_lab/ for complete resource management workflows integrated with laboratory operations.