Researchers at the University of British Columbia (UBC) have trained computers to predict the next design drug before it hits the market. This technology can save lives.

Law enforcement agencies are racing to identify and supervise new versions of dangerous psychotropic drugs, such as bath salts and synthetic opioids, although clandestine chemists are working to synthesize and distribute new molecules with the same mental effects as classic drugs of abuse.

It can take months to identify these so-called "legal stimulants" in seized pills or powders, during which thousands of people may have used a newly designed drug.

But new research is already helping law enforcement agencies around the world reduce the identification time from months to days, which is crucial in the race to identify and monitor new versions of dangerous psychotropic drugs.

“The vast majority of these design drugs have never been tested in humans and are completely unregulated. They are a major public health problem in emergency departments around the world,” said Dr. Michael Skinnider, a UBC medical student, who works at UBC’s Michael The Smith laboratory completed this research as a PhD student.

Skinnider and his colleagues used databases of known psychoactive substances provided by forensic laboratories around the world to train artificial intelligence algorithms to analyze the structure of these drugs. The algorithm they use is called a deep neural network, which is inspired by the structure and function of the human brain.

Based on this training, the model generated approximately 8.9 million potential design drugs.

After training these molecules and models, 196 newly designed drugs that appeared on the illegal market were tested. Researchers found that more than 90% of the information exists in the generation set.

In other words, the model can predict almost all new drugs discovered since training.

"The fact that we can predict which special drugs may appear on the market before they actually appear is a bit like the 2002 sci-fi movie "Minority Report", in which foresight of upcoming criminal activities will help to significantly reduce crime in the future In the world,” explained senior author Dr. David Wishart (him/him), Professor of Computational Science at the University of Alberta.

"In essence, our software provides law enforcement agencies and public health programs with an opportunity to target clandestine chemists and let them know what to look out for."

This still leaves the question of how to easily identify completely unknown substances.

The researchers found that the model also learned which molecules are more likely to appear on the market and which are less likely to appear. "We want to know if we can use this probability to determine what an unknown drug is-based solely on its quality-chemists can easily measure any pill or powder using mass spectrometry," said Dr. Leonard Foster (he/he), UBC Bio Professor of the Department of Chemistry, an internationally renowned expert in mass spectrometry.

The researchers tested this hypothesis using each of 196 newly designed drugs. Using quality alone, the researchers found that their model ranked the correct chemical structure of an unidentified design drug in the top 10 drug candidates 72% of the time. The integration of tandem mass spectrometry data (another easily available measurement result) increases it to 86%. When only one guess is involved, the model can predict the correct structure 51% of the time.

“We were shocked that the model performed so well, because it is generally considered an unsolvable problem to elucidate the entire chemical structure only through accurate mass measurements. Narrowing the list of billions of structures to 10 candidate structures can greatly Speed ​​up the speed at which chemists can identify newly designed drugs," Skinnyd points out.

Skinnider added that the same model can be used to discover a variety of new molecules, from identifying new performance-enhancing drugs for exercise stimulants to identifying previously unknown molecules in human blood and urine.

"Now there is a whole world of chemical'dark matter' outside our fingertips. I think the right artificial intelligence tools have a huge opportunity to illuminate this unknown chemical world," Skinnider said.

The UBC model is safely distributed by the New Psychoactive Substances Data Center, and is used by the U.S. Drug Enforcement Administration, the United Nations Office on Drugs and Crime, the European Drug and Drug Addiction Monitoring Center, and the German Federal Criminal Police Office.

The research "Deep generative model enables automatic structural clarification of new psychoactive substances" was published in "Natural Machine Intelligence".

-This press release was originally published on the University of British Columbia website

Tag: Artificial Intelligence Forensics Industry News Pharmaceutical Product Resources: Industry News

© 2021 Laboratory Manager. all rights reserved.