# Special Seminar in Computing & Mathematical Sciences

Thursday
November 21, 2013
4:00 PM

**Exploring chemical reaction spaces using a graph grammar approach.**

**Location:**Moore B280

During chemical transformations molecular entities can change their

quantitative physico-chemical properties while atom types or mass is

conserved. Furthermore, upon interaction, novel molecular species with

hitherto unknown physico-chemical properties may arise. Hence a formalism

for chemistry must be capable of capturing the algebraic and thermodynamic

structure of chemical processes. The formalization of chemical

transformation (reactions) as graph grammers shows the right level of

abstraction for the analysis of large and complex reaction networks. The

used double pushout formalism (DPO) guarantees important intrinsic

properties of chemical reaction systems such as mass conservation or

atom-to-atom mappings from educts to products. A graph grammar provides a

set of production rules (reactions) describing all possible graphs of the

given formal language and thuse is a compact representation of a chemical

reaction space. Iterative application of the production rules to a set of

starting graphs constructs a chemical reaction network that organizes the

molecular graphs in a hypergraph. Important questions such as the existance

and topology of complex reaction patterns, such as auto-catalysis, in large

reaction networks can be investigated in a systematic manner. The basic

idea here is to reformulate the chemical question as hyperflow problem and

to use optimization techniques such es integer linear programming (ILP) to

calculate solutions. Metabolic engineering, a sub-field of Synthetic

Biology, investigates the de-novo construction and design of enzyme

catalysed reaction networks for in vitro and/or in vivo production of

commodity chemicals. Since for many target molecules no natural pathways

are known, recombination of enzyme functionality under optimality criteria

becomes important to guide the design efforts. I illustrate how the

presented framework can be used in this research area to explore and rank

the entire network design space spanned by a set of enzymes, starting

and target compounds, and some measure of optimality.

Host: Stefan Badelt. Please contact me (badelt@caltech.edu) if you would like to meet with Christoph during his visit on Nov 21.

**Series**Computing + Mathematical Sciences Lecture Series

**Contact:** Lucinda Acosta at 626 395 4843/626 395 5707 lucinda@caltech.edu