(337c) The Confluence of Kinetic Modeling and Data Science | AIChE

(337c) The Confluence of Kinetic Modeling and Data Science

Authors 

Broadbelt, L. J. - Presenter, Northwestern University
Reaction pathway analysis and kinetic modeling are powerful tools to design novel routes to chemicals, identify optimal processing conditions, and suggest catalyst design strategies. We have developed methods for the assembly of kinetic models of substantive detail that link the atomic and process scales. We have applied our methodology to seemingly very disparate chemistries, yet applying a common methodology reveals that there are many ubiquitous features of complex reaction networks for chemical and biological systems. The first part of this talk will focus on mechanistic modeling of the conversion of hydrocarbons from renewable sources, from quantitative analysis of chemical catalysis by native inorganic constituents to mechanistic understanding of how enzymes achieve exquisite selectivity for similar conversion processes, leading to the potential for the design of novel (bio)chemical pathways. However, the design of novel pathways was carried out in the absence of any quantitative kinetic modeling, raising the intriguing question of whether data science approaches alone are sufficient to understand complex reaction networks. We demonstrate the application of data science methods to identify potential biopriviledged molecules, that is, molecules that are accessible from biological feedstocks and processes and serve as candidates for conversion to a full range of attractive products via selective chemical catalytic chemistries. Intriguingly, the ranking of potential biopriviledged molecules relies on unfurling potential reaction pathways from the candidate molecule to products of interest and calls for quantitative comparisons of kinetics, which demonstrates the confluence of kinetic modeling and data science and their symbiotic relationship