Network-like similarity graphs (NSGs), developed by Prof. J�rgen Bajorathand co-workers at University of Bonn, provide a framework for characterizing SAR landscapes in new ways. NSGs are used to visualize similarity and activity relationships within a compound data set. In NSGs, compounds are represented by nodes that are connected by an edge if the similarity between the corresponding compounds exceeds a predefined similarity threshold. Multiple-property based network like similarity graphs, (MP-NSGs), an extension of NSGs, provide a novel way of visualizing multiple-property information in an interactive and dynamic fashion that can be applied to a variety of visual data-mining and data-analysis aspects in design. The visualizations allow rapid understanding of what property (potency, selectivity, ADME) is in a smooth or rough landscape. Detailed local and global landscape analysis of compounds that have equal aggregate desirability scores can be detected using MP-NSGs and dissected using Spotfire DXP layouts to guide strategic decisions in medicinal chemistry programs. This presentation will highlight the use of MP-NSGs within Pfizer and illustrate with some examples the power of visual data-mining approaches to quickly assess pinch-points in multiple property space and evaluate strategic options to address optimization of multiple properties.

MP-NSGs: Multiple-property based network-like similarity graphs

246th National Meeting (2013)

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