Graph Mining Methods for Predictive Toxicology
Andreas Maunz · mediaTUM – the media and publications repository of the Technical University Munich (Technical University Munich) · 2013
Very efficient methods are required for analyzing the graph structures of molecules. A primary goal is to find subgraphs that occur primarily in the toxic or non-toxic compounds. However, the result sets are often too large for efficient post-processing. This work shows that more concise representations may be obtained efficiently, and that they can be of considerable utility for predictive models. A combination of structural and statistical constraints allows for efficient computation.