CultureMech: Microbial Culture Media Knowledge Graph

Overview

CultureMech is a comprehensive knowledge graph containing over 10,000 culture media recipes from major international repositories. It transforms unstructured media composition text from literature and laboratory records into standardized, machine-readable data through automated chemical entity extraction and ontology-based grounding.

The Challenge: Microbial cultivation protocols are scattered across scientific literature, culture collection databases, and laboratory notebooks in unstructured text formats. This makes it difficult to systematically analyze growth requirements, compare media compositions, or leverage this data for AI-driven predictions.

The Solution: CultureMech automatically extracts chemical entities from text-based media descriptions, parses concentrations, and grounds ingredients to standard chemical ontologies (ChEBI, PubChem), creating a unified knowledge graph that powers downstream AI tools.

Key Features

🧬 Comprehensive Coverage

🔬 Chemical Entity Extraction

🗂️ Ontology Grounding

📊 LinkML Schema

🌐 Browser-Based Exploration

Technical Architecture

Data Processing Pipeline

Raw Media Text
    ↓
Text Parsing & Cleaning
    ↓
Chemical Entity Recognition
    ↓
Concentration Extraction
    ↓
Ontology Mapping (ChEBI/PubChem)
    ↓
Structured Media Records
    ↓
CultureMech Knowledge Graph

Integration with CultureBotAI Ecosystem

CultureMech serves as the foundation of the AI curation pipeline:

Data Sources

CultureMech aggregates media recipes from:

Major Culture Collections

Scientific Literature

Standardized Media

Use Cases

1. Historical Data Mining

Extract cultivation conditions from decades of scientific literature to identify patterns in growth requirements across the microbial kingdom.

2. Media Standardization

Normalize media recipes from different sources to enable cross-institutional comparisons and meta-analyses.

3. AI Model Training

Provide structured training data for machine learning models that predict optimal growth conditions for novel organisms.

4. Comparative Analysis

Analyze relationships between taxonomic groups and their chemical growth requirements to inform cultivation strategies.

5. Novel Organism Cultivation

Leverage phylogenetic relationships and chemical similarity to recommend starting media for uncultivated organisms.

Example: Media Composition Extraction

Input (Unstructured Text):

"R2A medium containing (per liter): yeast extract (0.5 g),
proteose peptone (0.5 g), casamino acids (0.5 g),
glucose (0.5 g), soluble starch (0.5 g), K2HPO4 (0.3 g),
MgSO4·7H2O (0.05 g), sodium pyruvate (0.3 g), pH 7.2"

Output (Structured Data):

{
  "media_name": "R2A",
  "components": [
    {"compound": "yeast extract", "amount": 0.5, "unit": "g/L", "chebi_id": "CHEBI:82594"},
    {"compound": "proteose peptone", "amount": 0.5, "unit": "g/L"},
    {"compound": "casamino acids", "amount": 0.5, "unit": "g/L"},
    {"compound": "glucose", "amount": 0.5, "unit": "g/L", "chebi_id": "CHEBI:17234"},
    {"compound": "soluble starch", "amount": 0.5, "unit": "g/L", "chebi_id": "CHEBI:28017"},
    {"compound": "dipotassium phosphate", "amount": 0.3, "unit": "g/L", "chebi_id": "CHEBI:131527"},
    {"compound": "magnesium sulfate heptahydrate", "amount": 0.05, "unit": "g/L", "chebi_id": "CHEBI:31795"},
    {"compound": "sodium pyruvate", "amount": 0.3, "unit": "g/L", "chebi_id": "CHEBI:113958"}
  ],
  "ph": 7.2
}

Repository & Documentation

Topics

growth-media · microbes · microbial-ecology · microbiology · cultivation · media · media-ingredients · microbial-growth · microbial-culturing

Getting Started

Access the Data

  1. Browse Online: Visit the CultureMech web interface
  2. Download: Access structured data from the GitHub repository
  3. API Integration: Use with kg-microbe for programmatic access

Integration with Your Workflow

# Example: Loading CultureMech data
from culturemech import MediaKG

# Load the knowledge graph
kg = MediaKG()

# Search for media by organism
media = kg.search_by_organism("Escherichia coli")

# Get chemical composition
composition = kg.get_composition("LB medium")

# Export to standard format
kg.export(format="json", output="media_data.json")

Research Impact

CultureMech is part of the KG-Microbe knowledge graph ecosystem developed at Lawrence Berkeley National Laboratory. It supports:

Citation: See the KG-Microbe preprint for details on the broader knowledge graph ecosystem.

Contact & Collaboration

For questions about CultureMech or collaboration opportunities: