Researchers Test Digital Library of Over 1 Billion Fentanyl Analogs on Mock Pill

Researchers at Pacific Northwest National Laboratory reported a method that compares measurements from a suspicious pill to computer-generated reference values to identify fentanyl variants not yet cataloged by authorities. Bioanalytical chemist Tom Metz led the study.

The team used two customized instruments that detect chemical features shared by fentanyl compounds and distinguish unrelated molecules that share the same molecular mass.

Fentanyl is up to 100 times as potent as morphine. Two milligrams is potentially lethal. The United States reported more than 72,000 overdose deaths in 2023. Experts currently know about 60,000 of the billions of possible fentanyl forms.

Metz and colleagues computationally broke apart each of the roughly 60,000 known fentanyl and fentanyl-like molecules into fragments, then recombined the fragments to generate several billion hypothetical molecules. They removed nonsensical or implausible structures, such as those unlikely to penetrate the blood-brain barrier, and used machine learning to predict real-world chemical measurements for the remaining structures.

The final digital library contained over 1 billion fentanyl analogs.

The researchers created a mock pill containing traces of 12 commercially available fentanyl varieties, a chemically similar non-opioid decoy, and typical street ingredients such as caffeine. A blinded analytical chemist who had never seen the pill received the raw data and the computer-generated library and was asked to identify any fentanyl analogs present.

The chemist identified six of the mock pill’s fentanyl components perfectly and narrowed another four components down to a few possible candidates each.