Analyzing exemplified and Markush structures in patents

presentation · 6 years ago
by Christopher Kibbey, Jacquelyn L. Klug-McLeod, Bruce A. Lefker, Mark A. Mitchell, Robert Owen (Pfizer)
Medicinal chemists rely on patent intelligence at three distinct junctures of a research project: inception of an idea, ongoing competitive intelligence during project development, and preparation of the patent application. Typically, a medicinal chemist collaborates with expert information scientists and patent attorneys to evaluate competitor's intellectual property in order to assess where drug design efforts should be focused. The availability of comprehensive databases of exemplified and Markush structures make it feasible for medicinal chemists to carryout preliminary IP analyses on their own. The search capabilities of these databases generally are limited to substructure and similarity comparisons and few tools are capable of more in depth analysis of the chemical space disclosed within a patent. To address the need to provide medicinal chemists with improved tools for analyzing competitor’s IP, Pfizer has developed a suite of tools that provide convenient access to structures from patents and assess the overlap of chemical features in a set of patent structures with Pfizer compounds. In addition, we have developed an algorithm for performing a similarity search against a Markush structure that involves comparison of a reduced graph representation of a query structure with a set of reduced graphs generated from close-in enumeration of a Markush structure. Download slides