AI Systems Biology for Addiction: Large Scale Multi-Omics Network Modeling and AI Agents for Mechanistic Discovery
Date:
Invited Talk at Platform for Advanced Scientific Computing (PASC) 2025 Conference, Brugg, Switzerland
This talk presented novel advances in the integration of large-scale biological data using multiplex network models and AI-driven annotation strategies to uncover comorbid genetic pathways in addiction.
Topics covered:
- Integration of genomics, transcriptomics, proteomics, and metabolomics into multiplex networks using mixed integration strategies for downstream multiomic analysis.
- Generation of cell-type-specific predictive expression networks (scPENs) from single-cell RNA-seq data, using iRF-loop and RWRtoolkit.
- Construction of a 25-layer human prefrontal cortex multiplex network to understand opioid use disorder and smoking cessation overlap.
- Application of the MENTOR framework (Multiplex Embedding of Networks for Team-Based Omics Research) for gene clustering.
- Interpretation of gene clusters using bespoke LLM-powered agents to parse enriched biological functions, metabolic pathways, and signaling cascades.
- Future directions involving fine-tuned LLMs, SambaNova deployment, and potential clinical collaborations.
This talk demonstrated how scalable, AI-enhanced systems biology can guide insight into complex diseases by combining exascale computation, graph theory, and machine learning in biomedical research.