The valence state combination model: a generic framework for handling tautomers and protonation states

The consistent handling of molecules is probably the most basic and important requirement in the field of cheminformatics. Reliable results can only be obtained if the underlying calculations are independent of the specific way molecules are represented in the input data. However, ensuring consistency is a complex task with many pitfalls, an important one being the fact that the same molecule can be represented by different valence bond structures. In order to achieve reliability, a cheminformatics system needs to solve two fundamental problems. First, different choices of valence bond structures must be identified as the same molecule. Second, for each molecule all valence bond structures relevant to the context must be taken into consideration. The latter is especially important with regard to tautomers and protonation states, as these have considerable influence on physicochemical properties of molecules. We present a comprehensive method for the rapid and consistent generation of reasonable tautomers and protonation states for molecules relevant in the context of drug design. This method is based on a generic scheme, the Valence State Combination Model, which has been designed for the enumeration and scoring of valence bond structures in large data sets. In order to ensure our method's consistency, we have developed procedures which can serve as a general validation scheme for similar approaches. The analysis of both the average number of generated structures and the associated runtimes shows that our method is perfectly suited for typical cheminformatics applications. By comparison with frequently used and curated public data sets, we can demonstrate that the tautomers and protonation state produced by our method are chemically reasonable.