Theoretical Optimization of Enzymatic Biomass Processes
This dissertation introduces a complete, stochastically-based algorithmic framework Cellulect to study, optimize and predict hydrolysis processes of the structured biomass cellulose. The framework combines a comprehensive geometric model for the cellulosic substrate with microstructured crystalline...
|PDF Full Text
No Tags, Be the first to tag this record!
|This dissertation introduces a complete, stochastically-based algorithmic framework Cellulect to study, optimize and predict hydrolysis processes of the structured biomass cellulose.
The framework combines a comprehensive geometric model for the cellulosic substrate with microstructured crystalline/amorphous regions distribution, distinctive monomers, polymer chain lengths distribution and free surface area tracking. An efficient tracking algorithm, formulated in a serial fashion, performs the updates of the system. The updates take place reaction-wise. The notion of real time is preserved.
Advanced types of enzyme actions (random cuts, reduced/non-reduced end cuts, orientation, and the possibility of a fixed position of active centers) and their modular structure (carbohydrate-binding module with a flexible linker and a catalytic domain) are taken into account within the framework. The concept of state machines is adopted to model enzyme entities. This provides a reliable, powerful and maintainable approach for modelling already known enzyme features and can be extended with additional features not taken into account in the present work.
The provided extensive probabilistic catalytic mechanism description further includes adsorption, desorption, competitive inhibition by soluble product polymers, and dynamical bond-breaking reactions with inclusive dependence on monomers and their polymers states within the substrate. All incorporated parameters refer to specific system properties, providing a one to one relationship between degrees of freedom and available features of the model.
Finally, time propagation of the system is based on the modified stochastic Gillespie algorithm. It provides an exact stochastic time-reaction propagation algorithm, taking into account the random nature of reaction events as well as its random occurrences.
The framework is ready for constrained input parameter estimation with empirical data sets of product concentration profiles by utilizing common optimization routines. Verification of the available data for the most common enzyme kinds (EG, β-G, CBH) in the literature has been accomplished.
Sensitivity analysis of estimated model parameters were carried out. Dependency of various experimental input is shown. Optimization behavior in underdetermined conditions is inspected and visualized.
Results and predictions for mixtures of optimized enzymes, as well as a practical way to implement and utilize the Cellulect framework are also provided. The obtained results were compared to experimental literature data demonstrate the high flexibility, efficiency and accuracy of the presented framework for the prediction of the cellulose hydrolysis process.