Abstract Details

Poster 8: A novel approach to De Novo design utilising Reaction Networks

James E. A. Wallace1, 2, Beining Chen2, Michael J. Bodkin3, Val J. Gillet1
1Information School, Regent Court, 211 Portobello Street, Sheffield S1 4DP
2Dept. of Chemistry, University of Sheffield, Western Bank, Sheffield S10 2TN
3Eli Lilly, Erl Wood, United Kingdom
Information techniques can be used in the field of chemistry to synthesise potentially useful molecules in virtual space (so-called de novo design) that can then be assessed before being made for real, saving time and money (Hartenfeller and Schneider, 2011). Often with these techniques, the number of simulated molecules becomes so large that considering them all becomes impossible. Previous work within the iSchool (Hristozov et al., 2011, Patel et al., 2009) has created a method that takes the reactant and product of a chemical reaction, and records the changes as a ‘reaction vector’ that can be applied to other, similar structures to create new molecules. This method helps to ensure each simulated molecule is based on an existing reaction, making them more likely to be synthesisable in reality.
The reaction vector method has some limitations, however, when applied to real world situations, as in drug design many synthetic routes proceed over multiple steps rather than just one. Often the intermediates in a reaction sequence have low resemblance to either the starting molecule or the eventual product, making it difficult to determine if a sequence is likely to lead to a good product using the stepwise approach. The result is that many routes to useful products can be ignored.

This project attempts to find new solutions to the problem by utilising graph theory approaches to produce networks of reaction sequences by mining chemical databases. A reaction network can be considered as a graph in which molecules are connected via the reactions that can be used to transform one molecule into another. By connecting reactions in this manner, every path within the network will represent a synthetic route. The start and end molecules of these routes can then be used to generate a reaction vector as before, condensing the sequence to a single step and bypassing problems associated with intermediates.

HARTENFELLER, M. & SCHNEIDER, G. 2011. Enabling Future Drug Discovery by De Novo Design. Wiley Interdisciplinary Reviews: Computational Molecular Science, 1, 742-759.
HRISTOZOV, D., BODKIN, M., CHEN, B., PATEL, H. & GILLET, V. J. 2011. Validation of Reaction Vectors for De Novo Design. Library Design, Search Methods, and Applications of Fragment-Based Drug Design. American Chemical Society.
PATEL, H., BODKIN, M. J., CHEN, B. & GILLET, V. J. 2009. Knowledge-Based Approach to De Novo Design Using Reaction Vectors. Journal of Chemical Information and Modeling, 49, 1163-1184.

Return to Programme