University of Massachusetts Amherst uses PAC-MAN and AI to combat tuberculosis infections

Tuberculosis, caused by the bacterium Mycobacterium tuberculosis (Mtb), remains the world's deadliest single-agent infection, leading to 1.23 million deaths in 2024, according to the World Health Organization. On July 6, 2026, researchers from the University of Massachusetts Amherst unveiled innovative techniques that could significantly expedite the discovery of more effective tuberculosis treatments.

Advancements in Tuberculosis Drug Research

The research team has developed a pair of methods aimed at overcoming the formidable barrier posed by Mtb's unique outer cell membrane. Known as the mycomembrane, this structure is notoriously difficult for drugs, including antibiotics, to penetrate. The new approaches measure which chemical compounds can cross this barrier and utilize these measurements to predict the permeability of other compounds.

Sloan Siegrist, an associate professor of microbiology at UMass Amherst and a senior author of the study, emphasizes the uniqueness of Mtb: "Not only does it have two membranes that protect the cell from antimicrobial chemical compounds that we might use to kill it, its outer membrane is unlike any other biological barrier out there." This outer membrane contributes to Mtb's resilience against both the body's immune system and antibiotic treatments.

Introducing PAC-MAN for Enhanced Efficiency

In 2023, Siegrist collaborated with Marcos Pires, a chemistry professor at the University of Virginia, to introduce a technique called Peptidoglycan Accessibility Click-Mediated AssessmeNt, or PAC-MAN. This method allows researchers to test numerous compounds simultaneously, significantly improving the efficiency of drug discovery processes.

Despite the advancements offered by PAC-MAN, the team sought to enhance their approach further. They integrated computational biologists and chemists, including Anna Green from UMass Amherst's Manning College of Information and Computer Sciences, to harness known chemical measurements for predicting the compound uptake of unknown chemicals.

Machine Learning's Role in Drug Discovery

Anna Green's lab specializes in computational analysis of biological compounds. She notes, "Small molecules can be particularly difficult to analyze computationally. Because they come in all different sizes with a wide range of molecular connections, you can't describe them with a single measurement—by weight, say, or size." This complexity is where artificial intelligence becomes invaluable.

Green and her team developed a machine learning model, the Mycobacterial Permeability neural Network (MycoPermeNet), trained on data generated from PAC-MAN screenings. Once trained, the model can predict how well a compound can permeate the mycomembrane based solely on its chemical structure.

Using both PAC-MAN and MycoPermeNet, the researchers identified key attributes that determine how effectively a compound can penetrate the mycomembrane. They also discovered correlations between these features and a compound's ability to kill Mtb. Green stated, "The mycomembrane lets some molecules through and keeps others out. There must be something about this membrane, and about the chemistry of each molecule, that decides which ones get in—and our combined tools help us figure out which ones can get through, and why."