Tautomer enumeration and stability prediction for virtual screening on large chemical databases.

Tautomeric rearrangements affect the results of cheminformatics applications that depend on the knowledge of the 2D or 3D structure of a compound, such as tools for database searches, fingerprint generation, virtual screening, and physical-chemical properties prediction. In this paper we present TauThor, a tool to enumerate tautomers and predict tautomer stability in the aqueous medium. The enumeration is based on a recursive process that generates tautomers according to the general scheme HX-Y=Z left harpoon over right harpoon X=Y-ZH. The stability of a tautomer is calculated by using a library of 145 fragments associated with experimental tautomeric percentages in water and a pK(a) based-method that utilizes pK(a) values predicted by MoKa. Predicted tautomeric ratios based on pK(a) calculations were benchmarked against literature data for a set of eleven compounds. The FDA approved drugs database, the NCI database and two vendor databases - Specs Screening Library and Asinex Gold Collection - were used to illustrate the impact of tautomerism on chemical libraries and to evaluate the relative occurrences of alternative tautomeric forms.