(376d) Automated Reaction Mapping and Mechanism Identification

A reaction mapping is a one-to-one correspondence between atoms in the reactant and product of a chemical reaction. Such a mapping can be used to infer which bonds break and form and thus implies a possible reaction mechanism. In addition to providing mechanistic insight, reaction mappings are important for the calculation of chemical kinetics and the generation of transition state structures [1,2]. For the latter, where the three dimensional configuration plays a role, it is critical to consider the stereochemistry of the species involved.

Previous reaction mapping algorithms were typically based on maximum common subgraph methods that minimize the number of reaction centers, but this is not guaranteed to result in the optimal reaction mechanism. Crabtree and Mehta [3,4] developed a method based on canonical graph naming to minimize the number of bonds broken and formed; however, neither their method, nor any other reaction mapping approach in the literature, properly considers stereochemistry.

We have for the first time developed an automated computational method to generate stereochemically consistent reaction mappings [5]. Our approach is unique in that it is formulated as a mixed-integer linear optimization (MILP) model for which we have developed an efficient solution procedure. We have formulated two choices for the objective function: either minimize the number of bonds broken and formed or minimize the total bond order changes. Our method is demonstrated through several computational studies on chemical reaction databases, including GRI-Mech [6], KEGG LIGAND [7], and BioPath [8]. From over 8000 reactions tested, we were able to locate guaranteed optimal mappings for 99% within 1 CPU hour.

We have also developed a procedure to automatically detect molecular symmetries. By generating multiple reaction mappings and eliminating those equivalent by symmetry, we are able to identify all distinct mechanisms for a chemical reaction. This has never been previously achieved in the open literature, and is invaluable to researchers investigating possible chemical pathways. Both our reaction mapping method and mechanism determination approach are implemented in our web tool DREAM, which is freely available to the scientific community (http://selene.princeton.edu/dream/).

References:

[1] Westerberg, K. M.; Floudas, C. A. Locating all transition states and studying the reaction pathways of potential energy surfaces. J. Chem. Phys. 1999, 110, 9259-9295.

[2] Westerberg, K. M.; Floudas, C. A. Dynamics of Peptide Folding: Transition States and Reaction Pathways of Solvated and Unsolvated Tetra-Alanine. J. Global Optim. 1999, 15, 261-297.

[3] Crabtree, J. D.; Mehta, D. P. Automated reaction mapping. ACM J. Exp. Algor. 2009, 13, Article 1.15.

[4] Crabtree, J. D.; Mehta, D. P.; Kouri, T. M. An Open-Source Java Platform for Automated Reaction Mapping. J. Chem. Inf. Model. 2010, 50, 1751-1756.

[5] First, E. L.; Gounaris, C. E.; Floudas, C. A. Stereochemically Consistent Reaction Mapping and Identification of Multiple Reaction Mechanisms through Integer Linear Optimization. J. Chem. Inf. Model. 2012, 52, 84-92.

[6] Smith, G. P.; Golden, D. M.; Frenklach, M.; Moriarty, N. W.; Eiteneer, B.; Goldenberg, M.; Bowman, C. T.; Hanson, R. K.; Song, S.; Gardiner, Jr., W. C.; Lissianski, V. V.; Qin, Z. GRI-Mech 3.0. http://www.me.berkeley.edu/gri_mech/ (accessed Apr 18, 2011).

[7] Goto, S.; Okuno, Y.; Hattori, M.; Nishioka, T.; Kanehisa, M. LIGAND: database of chemical compounds and reactions in biological pathways. Nucleic Acids Res. 2002, 30, 402-404.

[8] Reitz, M.; Sacher, O.; Tarkhov, A.; Trümbach, D.; Gasteiger, J. Enabling the exploration of biochemical pathways. Org. Biomol. Chem. 2004, 2, 3226-3237.