Genotype-Phenotype Mapping Through Whole Cell Modelling

• Leaders: Jean (Jiayu) Wen, Marc Wilkins
• Location: Australian National University, Canberra 

Vision & Overarching Goal of Theme 2:
Theme 2 aims to make biology predictive by linking genotype → cell model → phenotype through data-driven and hybrid approaches that combine machine learning with mechanistic modelling. Theme 2 is MACSYS’s  genotype-to-phenotype modelling engine, turning genomic and multi-omics information into testable, quantitative predictions of cellular phenotypes. A key focus is explaining and predicting phenotypic variability, especially in complex, “ALL-gene” (omnigenic) settings, by developing effective whole-cell models that learn from large-scale data while retaining mechanistic interpretability. 

• REGISTER

Theme 2 is currently built around three integrated pillars: 

• Genomic large language / foundation models (gLLMs): learn reusable biological representations from DNA/RNA/protein sequences to uncover regulatory and functional rules. 
• Virtual Cell modelling: Integrate single-cell and multi-omics data to predict cell-state transitions and responses to perturbations, creating a direct bridge to experimental validation. 
• Hybrid modelling: combines mechanistic ODE/SDE frameworks with neural components (e.g., neural ODE/SDEs) and learned surrogates to capture whole-cell dynamics with improved interpretability and scalability. 

Aims of the Theme 2 Workshop:
To bring together MACSYS members and partners to: 

• align priorities across gLLMs, virtual cells, and hybrid whole-cell modelling, and identify high-value “bridge problems” that connect data-driven approaches to mechanistic whole-cell simulations 
• showcase emerging Theme 2 capabilities (e.g., sequence foundation models, perturbation prediction, virtual-cell state-transition modelling, and hybrid neural/mechanistic pipelines) and discuss how these accelerate genotype-to-phenotype prediction 
• strengthen integration with Theme 4 by focusing on how Theme 2 models can explain experimental phenotypes, predict responses to genetic and chemical perturbations, and guide biological discovery and biotechnology applications 
• define concrete next steps, including shared datasets and benchmarks, model-comparison protocols, cross-theme integration points, and opportunities for co-designed case studies with experimental and industry partners 

Structure of the Theme 2 Workshop:
Across two days, MACSYS members and partners will collaborate in person through short research spotlights, focused breakout discussions, and synthesis sessions aimed at defining actionable cross-theme research plans. Discussions will explicitly connect modelling advances to biological use-cases and experimental validation pathways. Where there is partner interest, a light-touch online component will be included to enable engagement from interstate and international partners.