The Evolution of Sonic Perception: From Digital Fragments to Shortwave Signals

Researchers at the University of Massachusetts Amherst have developed an artificial intelligence model called MycoPermeNet to accelerate the discovery of tuberculosis drugs. By combining this AI with a high-throughput screening technique known as PAC-MAN, the team can predict which chemical compounds can penetrate the bacterium’s unique mycomembrane, potentially speeding up the search for effective treatments for the deadly infection.

Why is the tuberculosis bacterium so resilient to antibiotics?

Tuberculosis, caused by the bacterium Mycobacterium tuberculosis (Mtb), remains the world’s deadliest single-agent caused infection. According to the World Health Organization, the disease was responsible for 1.23 million deaths in 2024.

The bacterium’s resilience stems from a unique outer cell membrane known as the mycomembrane. This biological barrier is exceptionally difficult for drugs and antibiotics to penetrate. Sloan Siegrist, an associate professor of microbiology at UMass Amherst, stated that the Mtb outer membrane is unlike any other biological barrier out there.

Because this membrane protects the cell from antimicrobial chemical compounds, finding effective treatments requires identifying specific “chinks” in the mycomembrane that allow drugs to pass through.

How does the PAC-MAN technique improve screening efficiency?

Historically, researchers had to test chemical compounds one at a time to determine if they could enter Mtb cells. This process is slow and limited by the vast number of possible chemical combinations.

In 2023, Siegrist and Marcos Pires, a professor of chemistry at the University of Virginia, introduced a technique called Peptidoglycan Accessibility Click-Mediated AssessmeNt (PAC-MAN). This method allows researchers to test many compounds in parallel, significantly increasing the speed of the screening process compared to traditional one-by-one testing.

What makes MycoPermeNet different from traditional drug testing?

While PAC-MAN increased testing speed, researchers sought a way to use data from known chemicals to predict how unknown chemicals would behave. This led to the development of MycoPermeNet, the Mycobacterial Permeability neural Network.

Developed by Anna Green, an assistant professor in UMass Amherst’s Manning College of Information and Computer Sciences, MycoPermeNet is a machine learning model trained on PAC-MAN screening data. The model identifies patterns in chemical structures to predict permeability.

How does machine learning predict molecular movement?

Small molecules are difficult to analyze through simple computation because they vary widely in size and molecular connections. A single measurement, such as weight, is insufficient to describe them.

According to Green, MycoPermeNet analyzes the chemical structure alone to predict how readily a compound permeates the mycomembrane. The model identifies specific physical properties and molecular substructures that allow a compound to bypass Mtb’s defenses. The research team found that these same features often correlate with a compound’s ability to kill the bacterium.

Who is involved in this research?

The study was published in the journal Nature Microbiology. The research was a collaborative effort between several institutions and was supported by the National Institutes of Health, the Gates Foundation, and the UMass Amherst Institute for Applied Life Sciences.

The research team included:

• Senior Authors: Sloan Siegrist (UMass Amherst), Anna Green (UMass Amherst), Marcos Pires (University of Virginia), Joel Freundlich (Rutgers University–New Jersey Medical School), and Wonpil Im (Lehigh University).
• Co-lead Authors: Irene Lepori (UMass Amherst), Nelson Evbarunegbe (UMass Amherst), Zichen Liu (University of Virginia), and Shasha Feng (Lehigh University).

Frequently Asked Questions

To stay updated on the latest breakthroughs in medical biotechnology and infectious disease research, subscribe to our newsletter or explore our related articles on AI in healthcare.