Reactions: Promise of AI in chemistry, and cannabis standards

Letters to the editor

Promise of AI in chemistry

The artificial intelligence–driven marriage between structural biology and drug development coupled with avant-garde illustrative programs can likely revolutionize biomedicine and human health by propelling investigations by structural biologists, chemists, and students. For example, the state-of-the-art ChimeraX, a next-generation molecular visualization program, renders biological molecules such as proteins, RNA, and complex structures with unprecedented clarity, speed, and exquisite detail with respect to the intra- and intermolecular interactions between their side chains, backbone, and ligands. Blender, a free and open-source 3D computer graphics software tool, ups the ante by permitting every student of structural biology to become a movie creator, cinematographer, and director through animation of the molecules, visual effects, backdrops, shadows, and mixing and matching myriad options to create stunning structural masterpieces. We have come a long way since Nobelists John Kendrew and Max Perutz stood in front of a gargantuan ball-and-stick myoglobin, allowing the world to visualize a protein.

By contrast to the days of yore when the arduous task of solving protein structures had parallels in drug discovery with a months-to-years span from initiation to completion or product realization, the resolution revolution in 2016, the release of AlphaFold in 2021, and more recently Meta AI (2022) have pushed frontiers in structural biology and propelled a renaissance in drug discovery—a natural corollary? Before the cryo-electron microscopy and computational, AI, and machine learning techniques matured, solving protein structures necessitated X-ray or nuclear magnetic resonance techniques and tried-and-tested tools such as hanging drop, molecular replacement, site-directed mutagenesis, hydrogen-deuterium exchange, and isotope labeling, while drug development relied on bottom-up synthesis, natural product screens, structure-activity relationships, and inhibition kinetics.

Today, the treasure trove of structures can retool drug discovery, which has transitioned into biologicals. The opportunity to revive small-molecule-based drugs through AI has never been more compelling, for it not only predicts protein structures but also catalyzes bottom-up and retrosynthesis; “invokes” appropriate reagents and catalysts; and can design drugs de novo or couple databases of known compounds with biomolecular structures to identify lead candidates. AI and machine learning make absorption, distribution, metabolism, and excretion and toxicology predictions and direct one-pot techniques to improve yields in candidates with reduced off-target hits while addressing scale-up and embracing green synthesis. The ability to impact human health, lifespan, and agriculture; achieve food security; clean water; and eradicate disease may perhaps be just around the bend with the freely available tools used à la folding at home.

Mahesh Narayan
El Paso, Texas

Cannabis standards

The US Pharmacopeia (USP) commends C&EN for highlighting the challenges in ensuring the quality and safety of cannabis products. The recent article draws needed attention to the issue of lab shopping for cannabis testing (Sept. 9, 2024, page 28). As an independent scientific organization developing quality standards for medicines and botanicals, we share concerns about the risks posed by the lack of uniform standards.

The USP has developed and published core quality attributes and tools that can help establish testing and quality benchmarks for cannabis testing (see J. Nat. Prod. 2020, DOI: 10.1021/acs.jnatprod.9b01200 and our article “Supporting the Quality of Cannabis for Medical Use”).

Our work sets forth specifications for nomenclature, identification criteria, strength and composition criteria, purity testing, and limits on contaminants and outlines robust, scientifically valid testing methods. When utilized—in laboratories or by incorporation into regulations—these tools provide transparency and validated methods for testing.

The article highlights several key issues that could be addressed by USP quality standards and resources:

▸ Lack of standardized testing methods

▸ Variability in results between laboratories

▸ Need for contaminant limits

▸ Education and training

We strongly urge all relevant parties to adopt rigorous, validated testing methods and quality standards. By implementing consistent public quality standards, we can help ensure the quality of cannabis products.

Jaap Venema
Rockville, Maryland

Editor’s note: Venema is the executive vice president and chief science officer of the USP.