Of the many approaches computational chemistry has brought to drug hunting, medicinal chemists often find matched molecular pair analysis (MMPA) and QSAR to be two of the most useful.<sup> </sup>The key strength of MMPA is in its interpretability, in contrast to other machine learning methods. Here we show how MMPA and QSAR may be combined to understand potency SAR while maintaining highly interpretable models. We further extend the paradigm by showing how network analysis can be applied to potency data sets to identify pivotal compounds and even in cases such as patents where the biological information is missing, useful inferences can be made. Insights from these analyses and outputs can be applied to current projects to provide new directions and with new results, models can be regenerated to close the loop and achieve the desired goal of learning machines that augment expert medicinal chemists.

Potency and patents, new arenas for matched molecular pair analysis in the artificial intelligence world

256th National Meeting (2018)

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