(591g) Introducing New Approaches to Gapfilling and Dynamic Flux Balances Analysis for Genome-Scale Models

Genome-scale Models (GSMs) of metabolism have become important tools for the silico study and design of metabolism in silico. The model reconstruction process typically involves collecting information from databases such as NCBI, UniProt, KEGG, ModelSeed, and KBase; however, incomplete systems knowledge leaves gaps in any genome-scale reconstruction. Current tools for addressing gaps, use databases of biochemical functionalities to address gaps on a per-metabolite basis and can provide multiple solutions. However, their major limitation is that they cannot avoid Thermodynamically Infeasible Cycles (TICs), invariably requiring lengthy manual curation. This is in part due to their per-metabolite approach. To address these limitations, an optimization-based multi-step method named OptFill is developed, which performs TIC-avoiding, whole-model (holistic) gapfilling. OptFill, as with other methods, uses a database of biochemical functionalities to address metabolic gaps, in contrast, it uses a three-step approach to maximize metabolites connected, minimize the number of reactions, and maximize the number of reversible reactions in each solution. Additionally, OptFill can be readily adapted to automate inherent TICs identification, aiding manual curation. OptFill was first applied to three fictional prokaryotic “toy” models of increasing sizes and to a published GSM of Escherichia coli, iJR904. This application resulted in holistic and infeasible cycle free gapfilling solutions. Following this, OptFill was applied to the reconstruction of a genome-scale model of Exophiala dermatitidis, iEde2091, which was used to study the cost of polyextremotolerance and the similarities of human and E. dermatitidis melanin synthesis. Overall, OptFill can address critical issues in automated development of high-quality GSMs.

As GSMs are under-defined systems of equations, optimization-based tools are required for their analysis, most frequently Flux Balance Analysis (FBA). An adaptation of FBA, dynamic FBA (dFBA), is a tool which allows for the study of a modeled system across time. Introduced here is a generalized Optimization- and explicit Runge-Kutta-based Approach (ORKA) to perform dFBA, which is more accurate and computationally tractable than existing approaches, namely the Static and Dynamic Optimization Approachs (SOA and DOA, respectively). A four-tissue (leaf, root, seed, and stem) model of Arabidopsis thaliana, p-ath773, is analyzed using ORKA. P-ath773 uniquely captures the core-metabolism of several stages of growth from seedling to senescence while strongly emphasizing plant-scale behavioral agreement between in silico results and in vivo data. Using ORKA, p-ath773 takes metabolic “snapshots” at hourly intervals throughout the lifecycle of an individual plant. This analysis shows the transition of metabolism and whole-plant growth, such as the evolution of sulfur metabolism and the diurnal flow of water throughout the plant. Specifically, p-ath773 shows how transpiration drives water flow through the plant and how water produced by leaf tissue metabolism may contribute significantly to transpired water. Investigation of sulfur metabolism shows frequent cross-compartment exchange of a standing pool of amino acids which is used to regulate proton flow. Additionally, p-ath773 has shown broad agreement with published plant-scale properties such as mass, maintenance, and senescence. Overall, p-ath773 serves as a scaffold for lifecycle models of other plants to further increase the range of hypotheses which can be investigated in silico.