Skip to content

research.md

Latest commit

 

History

History
187 lines (134 loc) · 10.9 KB

research.md

File metadata and controls

187 lines (134 loc) · 10.9 KB
Raw
layout page
title Research
description CultureBotAI research by Dr. Marcin P. Joachimiak focuses on AI-powered microbial cultivation, KG-Microbe knowledge graph development, and growth preference prediction using machine learning.
permalink /research/

Research Areas — AI and Machine Learning for Microbial Cultivation

CultureBotAI's research is led by Dr. Marcin P. Joachimiak at Lawrence Berkeley National Laboratory in Berkeley, California, focusing on artificial intelligence, machine learning, and microbiology.

CultureBotAI's research spans the intersection of artificial intelligence, machine learning, and microbiology, with a focus on transforming how we understand and manipulate microbial systems through the KG-Microbe knowledge graph.

Primary Research Areas

🦠 Cultivation of Isolated and Novel Organisms

We develop AI-powered approaches to successfully cultivate previously unculturable microorganisms. Our work focuses on:

  • Novel isolation techniques guided by machine learning predictions
  • Automated culture monitoring using computer vision and sensor networks
  • Optimization of growth media through iterative AI-driven experimentation
  • Scaling cultivation methods from lab bench to industrial applications

Key Challenges Addressed:

  • The "great plate count anomaly" - cultivating the 99% of microbes that resist standard cultivation
  • Identifying optimal growth conditions for fastidious organisms
  • Reducing time and resources required for successful cultivation

🔬 Culture Optimization Through Data-Driven Approaches

Our culture optimization research leverages big data and machine learning to dramatically improve cultivation success rates:

  • Environmental parameter optimization (temperature, pH, oxygen, nutrients)
  • Media composition prediction using computational approaches
  • Co-culture design for synthetic microbial communities

Technologies Employed:

  • High-throughput screening platforms
  • Automated liquid handling systems
  • Real-time monitoring sensors
  • Multi-objective optimization algorithms
  • Graph learning

🧠 Growth Preference Prediction Using AI/ML Methods

We develop sophisticated predictive models to understand and forecast microbial behavior:

Machine Learning Approaches

  • Deep neural networks for complex pattern recognition in microbial data
  • Gradient boosted decision trees for predictive modeling
  • Ensemble methods combining multiple predictive approaches

Data Integration:

  • Genomic and metagenomic sequences
  • Environmental metadata
  • Cultivation historical data
  • Literature-derived growth parameters

Knowledge Graph Development

KG-Microbe: Comprehensive Microbial Knowledge Integration

Our flagship KG-Microbe knowledge graph project developed by Dr. Marcin P. Joachimiak represents a breakthrough in microbial data integration:

  • Multi-source data integration from major biological databases
  • Ontology-driven organization ensuring semantic consistency
  • Machine-readable formats enabling automated reasoning
  • Community-driven updates ensuring data currency

Data Sources Integrated:

  • NCBI Taxonomy
  • UniProt protein databases
  • Environmental ontologies
  • Cultivation databases
  • Literature-derived facts

Current Projects

🔬 Automated Culture Monitoring Platform

Development of AI-powered systems for continuous culture monitoring using:

  • Iterative computational-experimental process
  • High-throughput cultivation
  • Physical parameter scanning

📊 Predictive Growth Modeling

Creating comprehensive models that predict:

  • Optimal growth conditions for target organisms
  • Media composition requirements
  • Co-culture compatibility

🌐 Knowledge Graph Applications

Expanding kg-microbe capabilities for:

  • Automated literature mining and fact extraction
  • Cross-organism growth condition prediction
  • Novel organism property inference
  • Integration with laboratory information systems

Tools Supporting Our Research

Our research goals are enabled by a suite of interconnected software tools:

For Cultivation of Novel Organisms

  • MicroGrowAgents - AI-driven media design with multi-agent reasoning
  • MicroGrowLink - Graph-based growth prediction using transformers (private repo, public release planned)
  • MATE-LLM - Literature protocol extraction with LLMs (private repo, public release planned)

For Culture Optimization

For Growth Preference Prediction

  • kg-microbe - Knowledge graph foundation with 864K+ validated species
  • KOGUT - Relational graph transformer for link prediction over kg-microbe (DOE CODE record; no public repository yet)
  • Explainable rule mining - Human-readable rules predicting cultivation media, published in CSBJ
  • assay-metadata - Phenotypic assay data from BacDive
  • eggnog_runner & eggnogtable - Genome functional annotation (eggnogtable is private, public release planned)
  • neurosymbolreason - Neurosymbolic analogy reasoning on knowledge graph embeddings

For Consortium Design & Community Modeling

  • CommunityMech - Microbial community interaction modeling with LinkML schema
  • PFASCommunityAgents - AI-driven consortium design for PFAS biodegradation (private repo, public release planned)

See all tools and workflows →

Collaborative Research

We actively collaborate with:

  • Academic institutions developing novel cultivation techniques
  • Industry partners scaling up microbial production processes
  • Government laboratories studying environmental microorganisms
  • Open source communities building computational biology tools

Publications & Preprints

Our methods are described in the KG-Microbe paper (GigaScience 2026), the explainable rule mining paper (CSBJ 2025), and the MicroGrowAgents preprint (bioRxiv 2026). See the Bibliography below, or the Publications page for the full list.

Get Involved

Interested in collaborating on microbial cultivation research? We welcome:

  • Research partnerships with academic and industry groups
  • Student researchers seeking challenging projects
  • Open source contributors to our software tools
  • Data contributors sharing cultivation datasets

Contact us to explore collaboration opportunities.

Frequently Asked Questions

What are CultureBotAI's main research areas?

CultureBotAI focuses on three main research areas: (1) cultivation of isolated and novel organisms using AI-powered approaches, (2) culture optimization through data-driven approaches and machine learning, and (3) growth preference prediction using AI/ML methods.

Who leads CultureBotAI's research?

CultureBotAI research is led by Dr. Marcin P. Joachimiak, a staff researcher at Lawrence Berkeley National Laboratory in Berkeley, California, specializing in microbiology, knowledge graph development, and computational biology.

What is the KG-Microbe knowledge graph?

KG-Microbe is a comprehensive modular knowledge graph developed by Dr. Marcin P. Joachimiak that integrates multi-source microbial data from major biological databases using ontology-driven organization to enable AI-driven insights and automated reasoning.

What machine learning methods does CultureBotAI use?

CultureBotAI employs deep neural networks for complex pattern recognition, gradient boosted decision trees for predictive modeling, ensemble methods combining multiple approaches, and graph learning for knowledge integration.

Where is CultureBotAI research conducted?

CultureBotAI research is conducted at Lawrence Berkeley National Laboratory in Berkeley, California, within the Environmental Genomics and Systems Biology Division, with collaborations at ABPDU and JBEI.

What are CultureBotAI's current projects?

Current projects include an automated culture monitoring platform using AI and high-throughput cultivation, predictive growth modeling for optimal conditions, and expanding kg-microbe capabilities for automated literature mining and cross-organism prediction.

Bibliography

  1. Santangelo BE, Hegde H, Caufield JH, Reese J, Kliegr T, Hunter LE, Lozupone CA, Mungall CJ, Joachimiak MP. KG-Microbe — Building Modular and Scalable Knowledge Graphs for Microbiome and Microbial Sciences. GigaScience. 2026;giag077. doi:10.1093/gigascience/giag077
  2. Máša P, Kliegr T, Joachimiak MP. Explainable rule-based prediction of cultivation media for microbes. Computational and Structural Biotechnology Journal. 2025;27:5194–5206. doi:10.1016/j.csbj.2025.10.014 · free full text
  3. Naseem S, Miller MA, Martinez-Gomez NC, Sun N, Joachimiak MP. MicroGrowAgents: An Agentic AI System for Microbial Cultivation Engineering. bioRxiv. 2026. doi:10.64898/2026.06.04.729985
  4. Joachimiak MP. Knowledge Oriented Graph Unified Transformer (KOGUT) v0.1 [software]. DOE CODE; 2025. doi:10.11578/dc.20260210.3 · DOE CODE 175162
  5. Caufield JH, Putman T, Schaper K, Unni DR, Hegde H, et al. (incl. Joachimiak MP). KG-Hub — building and exchanging biological knowledge graphs. Bioinformatics. 2023;39(7):btad418. doi:10.1093/bioinformatics/btad418 · free full text
  6. Caufield JH, Hegde H, Emonet V, Harris NL, Joachimiak MP, et al. Structured Prompt Interrogation and Recursive Extraction of Semantics (SPIRES): a method for populating knowledge bases using zero-shot learning. Bioinformatics. 2024;40(3):btae104. doi:10.1093/bioinformatics/btae104 · free full text {: .bibliography}

Full publication list →