DEVELOPMENT OF A MULTI-SCALE COMPUTATIONAL FRAMEWORK LINKING METABOLIC FLUX MODELS TO WHOLE-PLANT GROWTH AND BLSS PERFORMANCE

Overview & Synopsis

This dissertation developed, validated, and applied a novel multi-scale computational framework linking genome-scale metabolic models (GEMs) to functional-structural plant models (FSPMs) and bioregenerative life support system (BLSS) performance metrics for space agriculture optimization. The research addressed a critical gap in translating molecular-level metabolic predictions to system-level agricultural outcomes relevant to long-duration human spaceflight missions, where in situ food production is essential for crew sustenance and psychological well-being. The computational framework integrated constraint-based reconstruction and analysis (COBRA) methods at the metabolic scale with L-system-based architectural models at the whole-plant scale, connected through explicit scale-bridging mechanisms that capture source-sink dynamics and feedback signaling between biological organizational levels. Using publicly available data from PlantSEED, BRENDA enzyme database, TRY Plant Trait Database, and NASA controlled environment agriculture (CEA) publications, tissue-specific metabolic models were developed for four BLSS candidate crops: wheat, potato, rice, and lettuce. Validation against independent datasets including ¹³C metabolic flux analysis measurements, NASA Biomass Production Chamber growth trajectories, and EDEN ISS Antarctic greenhouse productivity data demonstrated high predictive accuracy (R² = 0.987 for total biomass, R² = 0.992 for grain mass) and confirmed that multi-scale integration improved prediction accuracy by 46% compared to single-scale modeling approaches, validating the central hypothesis that explicit cross-scale coupling enhances predictive capability. Global sensitivity analysis using Sobol variance decomposition identified 10 high-leverage metabolic parameters with disproportionate influence on harvest index, with ADP-glucose pyrophosphorylase (AGPase) activity exhibiting the greatest impact (S₁ = 0.182). In silico metabolic engineering simulations predicted that AGPase overexpression at twice baseline activity would increase harvest index by 18.7%, while synergistic combinations with photorespiratory bypass pathways achieved 31.4% improvement, exceeding additive predictions. Pareto optimization under realistic BLSS constraints identified non-intuitive intervention strategies that outperformed empirically-derived recommendations from single-scale studies. The framework predicted edible productivity of 12.8 g dry weight m⁻² day⁻¹ for wheat under optimal BLSS conditions and estimated cultivation area requirements of 50-80 m² per crew member for achieving caloric self-sufficiency on Mars surface missions. This research provides the first validated computational framework for rational prioritization of crop improvement strategies under space agriculture constraints, offering NASA, ESA, and international partners an evidence-based decision support tool for BLSS design, crop development, and mission planning. Keywords: systems biology, genome-scale metabolic models, functional-structural plant models, bioregenerative life support systems, flux balance analysis, multi-scale modeling, space agriculture