Formal Representation and Exploration of Inorganic Molecules with Graph-Theoretical Means

Abstract

We present the graph-based software Molassembler for organic and inorganic molecules. Molassembler provides algorithms for the construction of molecules built from any set of elements from the periodic table. In particular, poly-nuclear transition metal complexes and clusters can be considered. We motivate a molecular model in which structural information is encoded as a graph, and stereocenter configurations are represented as an abstract index of permutation for an arbitrary and freely extensible set of polyhedral shapes. A variety of shape classification methods are presented, devised and evaluated, yielding a reliable approach. The arrangements represented by abstract stereoconfiguration indices are filtered by three-dimensional feasibility, leading to practical stereodescriptors. An algorithm is presented to interpret stereocenter configurations from Cartesian coordinates. An extension of organic substituent ranking rules with the derived stereodescriptors expands its applicability to inorganic chemical space. The modification of the joint model composed of a graph and its stereocenter representations is discussed and an algorithm presented to prevent avoidable losses of chiral state through incurred shape, stereodescriptor, and ranking changes. Algorithms to compare molecules and canonicalize their representation are introduced. We show an approach to generate conformers with full stereoisomer control by four spatial dimension Distance Geometry. After discussing the model’s weaknesses, we demonstrate its strengths at some sample complexes. Finally, Molassembler is applied in the context of automated chemical reaction network exploration and the molecular design of a CO2-fixating complex.