(1338-C) Mastering complex challenges in controlled substance compliance using advanced Markush and similarity search technologies
Tuesday, February 6, 2024
12:00 PM – 1:00 PM EST
Location: Exhibit Halls AB
Abstract: The modern chemical and pharmaceutical industry faces demanding regulatory requirements related to the safe and legal use, storage and supply of compounds. The very first step towards meeting these legal requirements is the identification of controlled substances such as narcotics, psychotropics and chemical warfare agents. The size of molecular collections, encompassing both actual and virtual compound libraries, continues to grow, which necessitates robust cheminformatics tools for the efficient identification of controlled substances. The proliferation of synthetic drugs has led to the annual emergence of around 500 new psychoactive substances, prompting several countries to extend the coverage of their existing drug laws by defining generic control for groups of substances, rather than only listing individual drugs. These generic definitions are translated into a digital representation, typically in the form of Markush structures, representing a large or even unlimited number of structurally related substances under a single disclosure. Efficient search against Markush structures requires advanced methods that go well beyond a simple (sub)structure search. This poster will discuss a state-of-the-art query technology optimized for Markush search, and will showcase its application in Chemaxon’s Compliance Checker system. Another significant challenge for search engines in controlled compound compliance is the identification of substances that are "substantially similar" to controlled ones, as per the Federal Analogue Act (21 U.S.C. § 813). The notion of chemical similarity, a key concept in cheminformatics, is inherently subjective, leading to diverse interpretative approaches. This poster will highlight the challenges of defining chemical similarity in the context of compound compliance and the exploration of various techniques, from classic fingerprinting methods to machine learning. Finally, it will be shown that our similarity analysis approach resulted in a more accurate and cohesive similarity search technology, which we also implemented in the Compliance Checker application.