Molecular Design is often driven by the intuition of experienced medicinal chemists. When it comes to database searches, precise chemical queries, i.e. chemical patterns, have to be formulated. These patterns are the working horse of cheminformatics usually applied for filtering chemical libraries, defining functional groups, chemical space, and synthetic reactions. The most frequently used language for this task is SMARTS, a textual description with a complex syntax. The usage of SMARTS is non-trivial, especially for chemists which would like to only occasionally use these tools. In this talk, we present a novel software aiming at bridging this gap and enabling chemists to create SMARTS-based queries interactively. In order to make the difficult SMARTS syntax easily comprehensible, we developed a graphical representation of SMARTS, named SMARTSviewer and an interactive graphical editor named SMARTSeditor [1]. Both components are of great help in writing a chemical query. However, the pattern still has to be designed from scratch. Therefore, we recently developed the third component within this context, the SMARTSminer [2] supporting the medicinal chemist by automatically suggesting patterns as starting points for interactive design. The SMARTSminer is based on a contrast pattern algorithm specially tailored for chemical patterns. Starting from two sets of molecules A and B and two fractions a and b, the algorithm creates a list of chemical pattern able to separate these two sets. Each pattern ensures, that a% of molecules from set A are matched while at most b% of molecules from B are hit. All patterns fulfilling this criterion can be sorted by their support of separation as well as by their degree of generality. Whenever a pattern is picked and subsequently modified, the SMARTSminer displays the molecules hit and not hit in both sets A and B. In this way, the most frequent scenario for pattern generation is optimally supported while substantially reducing the technology barrier. In summary, the software enables chemists to transfer their ideas into chemical patterns that are then directly applicable in database searching. [1] Schomburg, K., Wetzer, L., Rarey, M. (2013). Interactive design of generic chemical patterns. Drug Discovery Today, 13:1-8. [2] Bietz, S., Schomburg, K. T., Hilbig, M., Rarey, M. (2015). Discriminative Chemical Patterns: Automatic and Interactive Design. Journal of Chemical Information and Modeling, 55(8):1535–1546.

Medicinal chemistry: From intuitive rules and examples to chemical patterns

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.