Recent advancements in metabolomics and fluxomics methods and ever-increasing amount of available experimental data necessitate systems-oriented methodologies for efficient and systematic integration of data into consistent large-scale kinetic models capable of giving new insights in cell physiology and providing reliable guidance for metabolic engineering. The development of large scale kinetic models is often hampered by (i) incomplete/no knowledge about rate laws in addition to uncertain or missing information about their kinetic parameters; (ii) difficulties in determining the exact intracellular states due to insufficient experimental information and large degrees of flexibility in the stoichiometry of metabolic networks. ORACLE (Optimization and Risk Analysis of Complex Living Entities), a computational framework that accounts explicitly for mechanistic properties of enzymes and integrates network thermodynamics, physico-chemical constraints in metabolic network and available experimental data can be used to address the aforementioned issues. ORACLE relies on Metabolic Control Analysis (MCA) and Thermodynamic Flux Balance Analysis (TFBA) supported by a range of statistics and systems theory methods, allowing us to uncover all the possible intracellular network states consistent with the available experimental data. This is followed by the computation of control coefficients in the metabolic network to guide the formulation of possible alternatives for metabolic engineering. We demonstrate the application of this methodology in an actual metabolic engineering problem: improving the xylose uptake rate of a glucose-xylose co-utilizing recombinant S. cerevisiae strain through a cycle between experimental lab work and modeling efforts. More specifically, using the given fermentation data we identified network flux and concentration profiles representing possible physiological states of the base strain. Then, we have identified targets that consistently lead to improved flux through xylose transporters (XTR) across these profiles. The targets identified by our kinetic models are experimentally validated demonstrating the predictive capabilities of ORACLE. Time: